Peter Burnhill

Type spaces

in-house norms in the typography of Aldus Manutius

Hyphen Press . London

 Contents

First published in book form by Hyphen Press, London, in 2003 Acknowledgements 7

This pdf file was made available free to download from hyphenpress.co.uk in 2019 Introduction 9
Text, drawings & diagrams copyright © Peter Burnhill 2003 Primary sources used in this enquiry 13
Note on methods of measurement 22
The book was designed by Peter Burnhill, Stafford, and Robin Kinross, London.
The text was typeset and made into pages by Robin Kinross in QuarkXPress.
This text was output in the typeface ITC Charter, designed by Matthew Carter. Typographic norms 25
Illustrations were scanned by Robin Kinross.
Aldine in-house norms 33
isbn 978 – 0– 907259 – 53 – 4

www.hyphenpress.co.uk Design analysis 1 Lascaris’s Erotemata and the octavo classics 71

2 Aldine character sets: relative proportions 89
For any reproduction in other publications, written permission must be sought from the
publisher
3 Typographic reference grids 101

The mould 122

Coda 124

Supplements 1 Positioning for meaning 125

2 Anthony Froshaug’s homage to Aldus 132
3 Ernest Hoch & Maurice Goldring, ‘Type size’ 136

Bibliography 140
Index 142

Postscript [2019] 144

For Ruth and Sonia who put up with ‘damn design’ for so long and so long
Acknowledgements

The need to ponder typographical detail in an extended and

relaxed way is not always possible in a rare-books reading room.
There, photocopying is not normally under the control of the stu-
dent. In addition, the handling of treasure to make detailed meas-
urements is, understandably, not always welcomed by library
staff. I am therefore grateful to Michael Twyman for the extended
loan of an Aldine Ovid. I have also inspected copies of Aldine edi-
tions in the British Library and Reading University Library. Many
thanks to James Mosley, former Librarian of the St Bride Printing
Library, London, for help through personal correspondence. I am
especially grateful to Graham Stevens and to Chris Butcher, the
typographer-librarian at Stafford College, for their invaluable help
in the preparation of computer-drawn reference grids and other
aids to measurement; and to my editor and typesetter Robin
Kinross, for his patience, criticisms, and good sense. Finally, my
thanks to colleagues Alan May, John Burgess, and John Cole, and
to past students and friends of the Design Department, Stafford
College, for their inspired and energetic boot-strapping over a
quarter of a century.

Ownership of books illustrated is given on p. 13, and elsewhere by

these codes:
bl: British Library
rul: Reading University Library
Thanks go to these two institutions for provision of the working
copies of the Aldine pages reproduced here.
Thanks also to: Walter Wilkes at the Technische Hochschule
Darmstadt, publisher of the edition of the Manuel Typographique,
from which the Fournier images have been taken; the estate of
Anthony Froshaug, for the copy of the homage to Aldus.

Type spaces 7

Systems of grouping and division are a more basic and durable
feature of a given metrological system than are the absolute values
of its measures.
Witold Kula, Measures and men, 1986
... it is not merely the shapes of the letters and the manner of making
them that requires study but also their size, the spacing between lines,
the size and proportion of the printed page ...
B. L. Ullman, The origin and development of humanistic script, 1960
To mention both typographic, and in the same breath/sentence,
grids, is strictly tautologous. The word typography means to
write/print using standard elements; to use standard elements
implies some modular relationship between such elements; since such
relationship is two-dimensional, it implies the determination of
dimensions which are both horizontal and vertical.
Anthony Froshaug, ‘Typography is a grid’, 1967
8 Type spaces
Introduction
The typography of Aldus Manutius has been of enduring interest
to designers and historians. In particular, the printed images of the
letters cut as punches, probably by Francesco Griffo of Bologna, to
the sizes called for by Aldus at Venice in the last decade of the fif-
teenth century, have been scrutinized by workers bent close over
letterform styles, their classification and attribution. Of these, the
Greek- and Latin-script cursive styles of letter, introduced by Aldus
as suitable for printing classical texts, have been subject to search
by historians for the specific styles of handwriting on which Griffo
may have been required by Aldus to ground his interpretations,
perhaps to endow them with authority as standards long before
such a notion became institutionalized.
As a typographer and teacher, not a historian, my interest in
Aldine typography lies less in the styles in which his letters were
dressed than in his use of spaces to convey visually the intrinsic
structure of the language. Although my pre-war art education had
had ‘The History of Styles’ as a required area of study, which
included Renaissance lettering, and calligraphy – as this had been
developed in the wake of the invention of printing from prefabri-
cated types and spaces – my first chance to look closely at the typo-
graphic construction of Aldine pages, free of the heavy breathing
of a rare-books reading room, came in 1957 when I found an
Aldine Seneca in a local second-hand bookshop.
I was attracted to the work, not by the italic characters in which
the book was composed, nor by the technical quality of the print-
ing, which was rather poor, but by the use of the simplest of all
possible means – one size of type – to create a close correspon-
dence between the hierarchical order of the work and the spaces
selected to define that structure at its differing levels: from the
space between words at the level of a phrase, to the grouping of
parts at the level of the document as a whole. Although the text
had obviously been planned with rubrication in mind, that work
had been left undone with no detriment to structural integrity.
Even though I could read neither Greek nor Latin, the one-to-one
relationship between the two dissimilar systems – the linguistic
and the spatial – was unmistakable. The job needed no additional
variables by way of colour or letterform sizes to help the reader
to chart a self-selected course through the seas of Seneca. The
9
Seneca edition served to overturn the cart on which the apples of
my pre-war art education had been piled, geometrically.
At the time of finding the Seneca, I had little experience of the
detail of the system of typographic measurement, such as the sub-
division of the type size – the em quadrat – associated with the
hand-composition of founder’s type, or the set values and spacing
sorts of mechanically engineered composition systems. As a conse-
quence, I was not at first curious about the actual structure of the
system of measurement – the typographic norms – used by Aldus
to orchestrate the labours of his editorial assistants, compositors
and pressmen; nor of the principles of spacing that had earned
him the accolade as bearer of the palm of printing, as this had been
awarded by the writing masters – the language teachers – of the
day (see the side-note). Furthermore, histories of printing tech-
nology had left the impression that printers – working before
the divorce of typefounding from the printing trade, to become a
separate industry – had contrived to put types and spaces together
without reference to a system of measurement.
Later, it became clear that Aldus must have commanded a very
refined system of dimensional control, long before manufacturers
of types and spaces for sale to the printing trade sought to regular-
ize type body sizes (mould sizes) on the grounds of providing
printers with more technically efficient work methods. It also
became clear that Aldus’s scale of dimensional values was a func-
tion of the type-casting system, not the consequence of the diktat
of some external ruler, such as the King’s foot, ‘Madame Guillo-
tine’, or some other unrelated unit of measurement determined for
their own use by a local guild of metalworkers. This is not to sug-
gest that the laterally adjustable mould with its sub-structures,
such as registers and stops for the positioning of matrices, came
into existence fully fledged, with no precursor in late-medieval or
early Renaissance technology. Nor is it to suggest that Aldus was
the father of this system. My guess is that in-house typographic
norms had been around since Gutenberg sorted mechanized script
into sub-sets by reference to common character widths – say, no
more than five or six groupings in all – then constructed a set of
fixed-width moulds to suit. If so, then the spacing sorts cast from
such moulds would vary in width, in correspondence with the sub-
sets of character width (a printing type being a space with a letter
cast on top).
It is noticeable that Gutenberg’s mechanical-looking letters
needed little if anything by way of side-bearings to set one letter
off from its neighbours in the context of a word. I suspect that the
10
‘Since up to now the palm in the art of
printing has, by general consent, been
awarded to Aldus Manutius, I will fol-
low him and state that the space he left
when the first five symbols are inserted
in the text, should be twice that which
we have left between words where no
punctuation mark intervenes. As for the
last symbol, you should leave a space
double that which we have given to the
other five symbols – or at least a third
more, as we find in the practice of Aldus
Manutius.’
Thus the writing master Juan de Yciar
with Juan de Vingles, writing around
1530 (‘Concerning the proportions
which should be observed when writ-
ing’). Here the printer, not the scribe, is
seen as the standard bearer. From Osley
(1980, p. 145).
Der Schriftgießer (The typefounder)
(Jost Amman, woodcut, 1568).
Carter (1969, p. 18) questions the relia-
bility of this as regards the ‘pyramidal
things’ on the shelf and in the caster’s
hand. Alan May, in conversation with
me, has tentatively suggested that these
could be fixed-width moulds: precur-
sors of the laterally variable mould.
However, more recent research by Stan
Nelson and others has now convinced
him that these ‘pyramidal things’ are
adjustable as to character width.
development of the more complex, laterally adjustable mould
with its attachments was in part a solution to such problems as the
need to provide slightly differing side-bearings for the rounder,
roman-styled lowercase characters, so as to achieve the snug
fitting required to define word images; and, possibly, for casting
combinations of two and three characters on a common body, as
commonly found in the Greek and Latin cursive character sets
made by Griffo at the call of Aldus. Certainly, the laterally
adjustable mould would have speeded up the casting of types.
Be all this as it may, what matters about the structure of Aldine
in-house norms was their combinatorial potential when used iso-
morphically by Aldus; and by his type designer, Griffo, to propor-
tion the primary dimensional attributes of the character sets he
cut for the firm, as these notes seek to show.
In the early days of printing, type sizes (mould sizes) were identi-
cal with the base-line to base-line (line-increment) of the job for
which the type had been specifically made to print, there being no
‘leading’ at the time for increasing the space between lines of con-
tinuous text, other than by casting types from a mould larger in its
body-sizing component than the mould for which the letters had
been dimensioned to fit in the first instance. Clearly, knowledge of
the proposed line-increment of a job, relative to the maximum per-
missible number of lines per page, was a precondition of type
design and punchcutting at this time in the development of typo-
graphy. It is one of the arguments of this essay into Aldine norms
that the line-increment (mould size) determined for the text pro-
vided the punchcutter with his scale of values for gauging the
grosser dimensional attributes and ratios of a projected set of
characters, without compromising judgement in the treatment of
lower order detail, such as the thickness of strokes and the treat-
ment of serifs.
Measurement indicates that Griffo used the same system of
dimensional reference for character face proportioning as that
used by Aldus to define the functional grouping of parts at higher
levels of linguistic order. These notes begin an examination of this
relationship as it may be gleaned from the pages that Griffo’s let-
ters were specifically sized to print. In other words, as distinct from
pages printed from his types as cast from moulds larger in the body
than the mould size / line-increment initially selected by Aldus for
the composition of the job in hand. New typefaces at this time
were made-to-measure.
Type spaces
Alan May comments:
‘Character fit was normally achieved by
adjusting the width of the matrix – by
filing if too wide or by “botching” if too
thin. Even after adjustable registers
were acquired, this was done so that
subsequent castings would match those
produced at the justifying stage.’
In a letter to me (13 February 1997)
James Mosley wrote: ‘I think you are
right. Leading of text matter for aes-
thetic reasons is largely an 18th-century
(and later) habit, and although there
are rare instances of line-spacing in
early printed books, apparently to bulk
out a short text, the spacing involved is
generally considerable, amounting to
the whole body size.
Cast leads are referred to by Pierre
Simon Fournier (Manuel typographique,
vol. 1, 1764), and also by T. C. Hansard
(Typographia, 1825). In the later 18th
century, the practice also developed of
casting types on a larger body than that
for which they were cut, Pica on
English, for example. My observations
of printing in earlier centuries suggests
that the founders generally cast types to
fit the body very tightly, and that print-
ers used them without leading. Some of
Plantin’s pages have ascenders touching
the descenders from the line above.
Sometimes, of course, a printer or
founder may not have had a mould to fit
a specific set of matrices, in which case
there may appear to be some line-
spacing.’
11
 With these constraints in mind, particular reference will be For a detailed account of the processes Primary sources used in this enquiry
made to the typography of the first and last of the jobs that Griffo entailed in cutting punches, see
Smeijers (1996).
did for Aldus Manutius before handing in his cards in 1502. These
were: a lowercase Greek running-hand for the Greek text of the
bilingual Greek grammar of Constantine Lascaris (the Erotemata), The following group of illustrations gives the reader a bird’s-eye-
as supplemented and edited by Aldus and manufactured by the view of the general appearance of the main objects of discussion,
printer, Andrea Torresani, in 1495; and the character sets needed namely:
to fit the thirty-two line page of the projected octavo classics series
which began publication in 1501, and which comprised Greek and Figures 1.1 – 1.3:
Latin running hands with a common set of Roman-styled capitals, Constantinus Lascaris, Erotemata, 1495 (quarto)
and a set of Arabic numerals. In addition, reference will be made British Library, c.2.91
to the control system of the renowned roman-styled Latins cut by
Griffo for Bembo’s De Aetna of 1495 /6, and to the second and third Figure 1.4:
of his Greeks. (The first and fourth Greeks being those of the Pietro Bembo, De Aetna, 1495 /6 (quarto)
Lascaris primer and the octavo classics series, respectively.) British Library, g.9289

My primary concern is to put the case that the Aldine typographic
norms provided a unified system of dimensional reference for use
on both axes of the page and at every level of order. Of secondary
interest to me is the notion that the mould sizes/line-increments
called for by Manutius following his first major experience of typo-
graphic design – the production of the Lascaris primer of 1495 –
were gauged by reference to a scale of dimensional values derived
from combinations of the spacing sorts of the Lascaris types. An
early venture of Aldus into printing, this primer also served to
introduce this grammarian – at the age of forty – to the innermost
secrets of the new technology: its measuring device, the laterally
variable mould.
I will also suggest that the mould used for the first casting of the
Lascaris types not only sired subsequent mould sizes called for by
Aldus but that it may also have fathered the name-designated type
sizes of subsequent centuries. In the early part of the sixteenth
century, throughout Europe, there was interest – to say nothing of
downright forgery – in both the literary content and the typo-
graphic syntax of Aldine publications, especially the forty-seven
titles of the standardized octavo series. Here are obvious grounds
for a link between the Aldine system and the system of type sizes
that appeared later.
Figures 1.5 – 1.8:
Ovid, Metamorphoses, 1502 (octavo)
library of Michael Twyman
Figures 1.9 – 1.14:
Seneca, Quaestiones naturales, libri vii, 1522 (large octavo)
author’s copy
All the pictures in this group are half real size (that is, 50 per cent
linear).
The term ‘black art’ is often used to refer
to secrecy in the methods used by the
early printers. Mathematics was also
known as the ‘black art’ associated with
the devil’s work, as evidenced by the
mathematician’s ability to predict the
future in respect of such events as an
eclipse. At a more mundane level, the
sliding mechanism of the two halves of
the printer’s type-casting mould for pro-
portioning the widths of spacing mater-
ial and the various widths of Latin let-
ter, is not so very different in principle
from the short scale sliding on a gradu-
ated scale to give fractional readings, as
invented by the Burgundian, P. Vernier
(c. 1580– 1637). I am grateful to my col-
league Alan May for this observation.

12 Type spaces 13
1 Primary sources: Lascaris, Erotemata (1.1 – 1.3); Bembo, De Aetna (1.4)

1.1 1.3

1.2 1.4
1 Primary sources: Ovid, Metamorphoses

1.5 1.7

1.6 1.8

16 Type spaces 17
1 Primary sources: Seneca, Quaestiones naturales

1.9 1.11

1.10 1.12
1 Primary sources: Seneca, Quaestiones naturales

1.13

1.14
Note on methods of measurement Computer-drawn base-lines and machine calculation of
line-increments (mould sizes)

The measurements of printed material presented in this book have Lascaris, Erotemata, 1495.
been made in this way: First casting.
Line-increment: 6.15 mm
1 Produce photocopy enlargements (say, 8 ×) of primary source + 2.0% = 6.27
material. or + 2.5% = 6.30
2 Using the enlargements, mark off the distance measured verti- nearest 0.1 mm = 6.3
cally between consecutive lines of text at a point on the lines
where characters (say, m or x) appear one above the other.
3 Divide the line-to-line interval found at 2 into twelve equal
parts.
4 Construct transparent fine line reference grids with x and y
co-ordinates set by reference to the twelfth part of the line-to-line
increment found at 3.
5 Use the reference grids as overlays to check the veracity of the De Aetna, roman, 1495 /6.
notion that the Aldine system of dimensional reference was based Line-increment: 5.62 mm
+ 2.0% = 5.73
on combining multiples of the twelfth part of the mould size of the or + 2.5% = 5.80
type in question. nearest 0.1 mm = 5.8

I also use an eye-glass (10 ×) reticulated at 0.1 mm intervals, as
used for inspecting circuitry in the manufacture of computer
chips. Another useful aid is a zebra-striped strip of acetate with
intervals corresponding to some multiple of one-twelfth part of
the mould size / line-increment of the type in question.
‘Peak’, 10 × de-lux, made by Bishop
Graphics Inc. (‘Manufacturers of
Engineering and Drafting Aids’), West
Lake Village, California, ca 91359, usa

When the purpose of measurement is not to determine absolutes,

such as mould sizes, but ratios as these fall on the retina of the
reader, such as x-height : inter-x-height space, then such incidental Seneca, italic, 1522.
variables as paper shrinkage and cockle are irrelevant. Otherwise, Line-increment: 3.92 mm
+ 2.0% = 3.998
some 2 per cent is allowed for incidentals. (See opposite.) or + 2.5% = 4.017
nearest 0.1 mm = 4.0

22 Type spaces 23
Typographic norms

The widths of the spacing materials of our erstwhile Anglo-

American and Didot systems of typographic measurement are well
known to an older generation of present-day typographers and
printing-trade workers, and were set clear to view by Anthony
Froshaug (1964). But historians of typography have not been very
interested in the syntactic function of spaces. Historians and com-
mentators have certainly been interested in the development of
the printer’s mould. But no one has yet investigated it as a device
for establishing a scale of spatial values. Of the Anglo-American
norms, Southward, in his technical manual Modern printing: a
handbook (1912, p. 112), says:
‘Spaces, in width, are aliquot parts of the em unit of a fount.
There are five spaces generally supplied [by type founders] – the
en quad ... the thick space, the middle space, the thin space, and
the hair space, thus:
En quad = 1⁄2 of the em, or 2 to em;
Thick space = 1⁄3 of the em, or 3 to em;
Middle space = 1⁄4 of the em, or 4 to em;
Thin space = 1⁄5 of the em, or 5 to em.
The hair space varies according to the size of the body. In the
smaller founts it is from 1⁄6 to 1⁄10 of the em; in some pica founts
it is 1⁄12 .
The thick, middle, and thin spaces are spoken of respectively
as the 3, 4, and 5 to em spaces.
These spaces can, in combination, be used to fill up any
space with sufficient exactitude. It is calculated that in pica no line
need be longer than another by more than 1⁄360 of an inch, or in
nonpareil half of that, viz. 1⁄720 , if the spaces are properly com-
bined.
The following table shows how the spaces between the
single thick and two thick spaces may be graduated:
One remove more, in width, two thin spaces;
one remove again, in width, a thin and middle;
one remove again, in width, the en quadrat;
one remove again, in width, thick and thin;
one remove again, in width, thick and middle;
one remove again, in width, two thick spaces.
If the compositor wishes to lessen the space from the thick
space downward he may try first the middle space, then the thin
space. The hair space, as already mentioned, is to be used only
under exceptional circumstances.’

Type spaces 25
For Southward, ‘sufficient exactitude’ seems limited to a concern
for differing spacing widths as technical means for tightening up
(the justification) of lines of metal type to a fixed measure, by
altering the space between words, rather than the use of intervals
for grouping parts in a wider, linguistic context – and in this
Southward is a representative voice of the printing trade of his
time. Strict economy in the use of costly paper also loomed large
in the practice of the early printer, often resulting in strategies
which, given time, assumed the status of immutable styles, such
as visually ‘justified’ pages.
Of earlier times, Davis & Carter, in a footnote to Moxon’s
Mechanick exercises (1962, p. 103), say: ‘The present convention
for the thickness of spaces (thick, 3 to the em; mid, 4 to the em;
thin, 5 to the em) is of uncertain age. Johnson’s Typographia (1824,
p. 101) shows that room was found in the case for three spaces
when the long s went out of use. Jacobi’s Printing (1890, p. 21)
gives their present value. Fertel (1723, p. 18), advises the printer to
ask the typefounder to cast him spaces of three different widths
“so that the journeyman may justify his lines more accurately and
easily”; but he contemplates the use of only two as the common
practice (pp. 16, 17). Fournier, (1764, p. 225) writes of spaces of
four or five different thicknesses; and the same author in the
Encyclopédie (‘Imprimerie en Caractères’) says “five or six thick-
nesses”, perhaps reckoning en- and em-quadrats among them.’
So Fertel too, in the eighteenth century, focuses on justifica-
tion as the principal reason for wanting spaces of differing width.
He wants them for technical reasons, as one might expect of the
printer who has to ensure that three-dimensional, movable metal
types will lock-up securely in a rectangular iron chase, for transfer
to the bed of a printing press. Here is a point of tension between
the meaning of a text and the technology of its realization: a long-
running point of possible difference between editors and typo-
graphers, concerned with the linguistic structure of a text, and
printers concerned with the mechanics of its production.
The earliest account of the use of spacing sorts as measuring
devices for use in the tooling-up of face gauges in preparation for
cutting the punches for a set of characters is probably that of
Moxon (1683 – 4). He writes of using seven ‘thin’ spaces. For him,
the thin space was one-seventh part of the projected type-body
size, with each seventh divided into six equal parts – ‘by curious
working’ – to produce a scale of 42 parts to the body of the project-
ed typeface, for use in the drafting and the fashioning of alphabets
(figure 2). Fournier (1764) also describes the division of the body
26
Although the term ‘justification’ is now
used to refer to lines of prose set to the
same length by arbitrarily changing the
space between the words, for letter-
press all lines had to be justified – made
tight in the stick – including lines of
verse, or other text, that were not of
equal length. Otherwise the type would
fall apart when taken out of the stick;
that is, become ‘pied’.
The major work of formalizing a system
of typographic measurement was
begun in France at the end of the seven-
teenth century, under the auspices of
the Académie des Sciences. See Mosley
(1991, 1997) for detailed discussion of
this project.
into seven equal parts (figure 3), as seemingly did William Caslon,
three parts being allocated to the x-height, with the remaining
four being shared equally between the upper and lower storeys of
the three-storey Latin-script writing system.
Closer in time to Francesco Griffo (d. 1518) is the method
described by the polymath Gerardus Mercator (1512 – 94), who
divides the square of the lowercase x-height of his Latin minus-
cules into twelve equal parts (figure 4) for proportioning the vari-
ous widths of his italic-styled characters, the angle of slope being
the diagonal of a rectangle with sides in the ratio of 1 : 12. His
extenders and capitals are allowed to spring freely, but with con-
trolled élan, beyond the confines of his modulus. These exemplars,
however, were for engraving letters for the captioning of maps
printed intaglio, and as such were unconstrained by the limits of
the rectangular and infinitely variable width of a printer’s type-
casting mould.
In his discussion of the methods of Mercator, Osley (1980,
p. 189) writes: ‘It is ironic that, whereas many Italian masters
prided themselves on their “geometrical” construction of letters,
Mercator, a highly trained mathematician, never mentions the
subject’. Another feature of Mercator’s approach is the notion of
assembling the form of a letter from a very limited set of one-
stroke elements. In the visual structure of written language, these
elements function at a level lower than that of the whole letter,
and they involve the concept of combining subsets of a universal
set, just as we find in the application of set theory. It is interesting
to contrast the elemental approach of Mercator to the construction
of a letter with that of Moxon whose outline characters and rule-
and-compass way of describing the shape of a letter provides little
sense of the structure of letterforms. Griffo, for his part, wrote Structure: the unifying principle that
nothing, leaving us to glean his methods and his understanding as governs the relationship between the
parts of a whole. In typography, this is
best we may from what exists on the printed page. at once both technical (dimensional)
and linguistic.
The illustrations on the following pages show the gauging systems
discussed by Moxon and Fournier respectively, and something of
the principles used by Mercator, as he describes in his treatise
Literarum Latinarum (1540).
Type spaces 27
2 Moxon’s scale 3 Fournier’s system

In this case, an em quadrat divided into From: Fournier (1764, vol. 3, Fournier
42 parts is used as a scale for gauging on typefounding, p. 23).
the dimensions of large character-sets. Note the footnotes by Harry Carter. The
For smaller faces, Moxon writes of ‘plate viii’ that he refers to must be
dividing his ‘thin’ space (one-seventh plate iii, which is reproduced below.
part of the em quadrat) into fewer than
6 parts for face-gauging (say, only 2
parts), making a scale of only 14 parts to
the body (em quad), compared with the
42 parts to the body as shown here for
large faces.

Below, from: Fournier (1764, Manuel

typographique, vol. 1, plates at rear).
‘Instruments used for punchcutting’.
A face-gauge for roman and italic is
shown top left, at 1.

From: Moxon (1683 – 4, pp. 124, 125).

28
4 Mercator’s system

From: Osley (1980, pp. 194, 200).

30
Aldine in-house norms

I was made curious about the detail of Aldine in-house norms by

the interval that fell between Griffo’s roman Q and the cursive u of
that ubiquitous, long-tailed, single-bodied sort as I saw it in my
large-octavo copy of an Aldine annotated Seneca of 1522. This
book was composed in the Greek and Latin character sets and the
Arabic numerals originally made by Griffo around 1500 for the
octavo classics series. In those books the requirement was a page
of thirty-two lines maximum, with an em quadrat / line-increment
of some 4.0 mm. When trimmed, the size of the page was about
160 × 100 mm.

Griffo’s gap in the Seneca

(real size / 800%).

Left: From the Seneca (sig. p1r, 100%).

This shows the Greek and Latin cursives
made by Griffo for the octavo classics
series. Note the ‘hanging indents’. Note
too the size of the capitals: this allows
them to settle comfortably between
lines and still function well as clear
markers of meaningful change.

In the Seneca there seemed to be no good reason for the gap

between the Q and the u. However, when looking at an early mem-
ber of the octavo classic series – Ovid’s Metamorphoses of 1502 –
the reason became silly-obvious. Griffo had done it to play in with
the practice – carried over from the handwritten mode of charac-
ter assembly to the typographic, prefabricated mode – of using a
stand-off capital: uniformed in the square-cut, stony-faced author-
ity of Imperial Rome to pull rank over cohorts of Greek and Latin
lower orders. By 1522, with Aldus seven years dead and the reput-
edly tight-fisted Torresani in the saddle, the other ranks had been
cuddled-up to the imperious sentinel to form a ‘hanging indent’.
Much later, these would be reduced by the mechanics of the man- Leto’s hand, c. 1470 (reduced).

Type spaces 33
ual typewriter to an all-purpose, knee-jerk indent; or, much worse,
swallowed without trace by the encroaching entropy of mid-twen-
tieth-century formalism.
My curiosity now switched from Griffo’s gap to the subtly
changing duration of the intervals following the stand-off mark-
ers. It was clear that these were made to change in counter-step
with the varying chest measurements – from I to M – of the
Imperial Guard so as to maintain as nearly as possible the vertical
axis of the left-ranged lower orders. This would not have been a
difficult manoeuvre in the handwritten mode of letter assembly –
if the eye is kept on a pre-scribed reference grid. But in the typo-
graphic mode it would have needed to be strictly choreographed:
in the absence of house rules, compositors working on the same
job might see their setting, when combined, ending in inconsist-
ency. As even a cursory glance at Aldine pages shows, Aldus had a
passion for structural clarity. It is not difficult to see when one of
the team had been too long at the stoup; or, being untutored, had
stuck the same space through the lot to make a dog’s hind leg of it.
Subtlety in the changing duration of the intervals following the
stand-off markers, which appear at every turn of the page, sug-
gested to me that a fine-grained set of spacing sorts was in use: as
fine as, if not finer than, the structure of the old Anglo-American
34
From the Ovid, 1502 (100%).
Master addressing a compositor, while
counting on his fingers. Note the ser-
vant, centre back, carrying in a stoup of
ale to slake the thirst of workers. Lead
poisoning, caused by handling type and
by lead dust in typecases, was for long a
scourge of the printing trade. Probably
engraved by Moses Thym, a Wittenberg
printer, working around 1613.
(Simpson, 1935.)
system of sub-modular control. And so it was. But then this was
not to be wondered at, given the practice of very close word-
spacing carried over from the late-medieval page to the reformed
page of the humanists.
Measurement of the widths of the printed images of rising
spacing sorts, as these can be found without much difficulty in the
composition of the Seneca of 1522 and the octavo edition of Ovid’s
Metamorphoses of 1502, indicates the use of spaces as fine as one-
twelfth part of the line-increment (type body size). Our Anglo-
American ‘thin’ space (5 to the em, or 2.4 twelfths) appears not to
form part of the Aldine system of sub-modular control. The Aldine
equivalent of the ‘thin’ space is the sixth part of unity (2 twelfths of
the em quadrat). This preserves the factorially rich duodecimal
structure of the Aldine standard, with its division of the em
quadrat by 2 as many as five times, and by 3, once. Thus:
12 6 4 3 2 1.5 1
From Erotemata (sig. l4, 100%). Notice
The nearest Anglo-American equivalent of the sub-modular
the close word-spacing; also the rising
twelfth of the Aldine system is the ‘hair’ space which, as men- word spaces on lines 5, 6, 9, & 14. The
tioned by Southward (1912, p. 112), varied in width in his day last of these starts just below the base-
depending on the type size. Normally reserved for the tightening line, hinting that the types were cast
very low in the mould aperture, with
up of slackly composed lines of metal type to a predetermined down-kerned descenders. See also
measure in the composing stick (justification) and for letter- figure 7.
spacing by eye of all-capital character groups, the Anglo-American
‘hair’ space has not been used recently for the calculation of inter-
vals. In contrast, the sub-modular twelfth of the Aldine system was
used as the basic unit of measurement for gauging intervals on
both axes of the page of less than em quadrat / type size, including
character image sizes, and not simply as a word space or a compo-
nent of word spaces.
Using Southward’s terminology, the Aldine typographic norms
can be expressed as:
en quadrat = 1⁄2 of the em, or 2 to em, or 6 twelfths;
thick space = 1⁄3 of the em, or 3 to em, or 4 twelfths;
middle space = 1⁄4 of the em, or 4 to em, or 3 twelfths;
thin space = 1⁄6 of the em, or 6 to em, or 2 twelfths;
v. thin space = 1⁄8 of the em, or 8 to em, or 1.5 twelfths;
hair space = 1⁄12 of the em, or twelve to em.
Type spaces 35
5 Typographic norms
Typographic norms:
above, the Aldine norms;
below, the Anglo-American norms of
the nineteenth century.
Schematic representation of the
adjustable mould.
a: aperture
b: body-sizing components
For the printer’s mould as a sliding
mechanism built of finely honed parts,
see May (1995 /6).
Teaching aid developed by Alan May
and John Burgess at Stafford College,
c. 1970, for connecting the concepts
‘mould aperture’, ‘set value’ and
‘character field’. The black spaces
between the character fields are invert-
ed square-set letter Ms. The interline
black is printed from 12 pt face rules.
By combining these spacing sorts – these deliberately engin-
eered fractions of the modulus – in different ways, a finely graded
scale of dimensional reference could be created, thus:
1⁄ = hair space
12
2⁄ = thin space (1 ⁄6)
12
3⁄ = middle space (1⁄4)
12
4⁄
12 = thick space (1⁄3)
5⁄ = middle space + thin space
12
6⁄ = en space (1⁄2)
12
7⁄ = thick space + middle space
12
8⁄ = two thick spaces (2⁄3)
12
9⁄ = en space + middle space (3⁄4)
12
10⁄ = en space + thick space (5⁄6)
12
11 ⁄
12 = two thick spaces + one middle space
12 ⁄
12 = two en spaces or one em quadrat
The addition of a space of 1.5 of the em quadrat – that is, one half of
the middle space, or one eighth of the em quad – would provide a
basic increment of 1⁄24 of the type body size. Spaces of 1.5 of the em
quadrat can be found in the composition of Aldine pages. Such a
scale could have been used not only for face-gauging, but also for
sizing the parts of moulds which determine type body sizes. (See
‘The mould’, p. 122.)
—
The following illustrations (figures 6 – 19) show pages from the
Aldine publications under discussion, together with annotated
enlargements of the printed images of spacing sorts to be found
there. Although a full range of spacing sorts does not show up in
printed image form in the copies of the Bembo and the Lascaris
primer that I have examined, I have no reason to think that the
spacing system used in these books would be any different in
structure from that perceivable as printed images in the Ovid and
the Seneca. In all instances, the body size appears to be divided
into twelve equal parts for character face gauging, as distinct from
Fournier’s basic seven parts. Once the division of the body into See p. 22 for a note on methods of
twelve parts has been established, it is not difficult to measure the measurement.
widths of intervals without recourse to images of risen spaces.
It is also clear that Fournier’s seventh is disjoint relative to a
duodecimal system of measurement. (Could this be the source of
the anomaly, which today characterizes our inability to specify
character image size, as distinct from type size, in typographic
design practice?)
Type spaces 37
6 Lascaris, Erotemata, first casting, sig. a 3r

400%

The low base-line provides sufficient space above the capitals for See figure 21, for evidence of the
the housing of diacritical marks, when these are required to be low base-line (writing line) of the
Lascaris types.
shown above the capitals of the Greek, as was traditional. This
caused the descending strokes of characters in both scripts to be
down-kerned. Note too the differently styled capital ‘I’s (top,
Latin; lower, Greek).

100%

38
7 Lascaris, Erotemata, first casting, sig. l 4r

400%

100%

40
 8 Bembo, De Aetna

400%

100%
Hair space is one twelfth of body size; x-height is centred on
the body.

42 Type spaces 43
100%
9 Ovid, Metamorphoses, sig. q1v–q2

800%
10 Seneca, Quaestiones naturales, sig. e2r

800%

The interlinear space (6 units) was probably created by setting

en quads sideways.

100%

46
11 Seneca, Quaestiones naturales, sig. m1r

800%

100%

Note that, within a paragraph, matter to be discussed is preceded

48
by a large space. Such large spaces could not have been used as
markers if word-spacing had been erratic.
 12 Seneca, Quaestiones naturales, sig. g5v

800%

100%

Type spaces 51
13 Seneca, Quaestiones naturales, sig. l7r

800%

Note the size of the punctuation marks, and compare with the
visual weakness of present-day punctuation.

100%

52
14 Seneca, Quaestiones naturales, sig. o2r

800%

100%

54
 15 Seneca, Quaestiones naturales, sig. f5r

800%

100%

Type spaces 57
16 Seneca, Quaestiones naturales, sig. q6r

800%

100%

58
17 Seneca, Quaestiones naturales, sig. g7r

800%

100%

60
18 Seneca, Quaestiones naturales, sig. a7r

800%

The image of a thick space of 4 units – one third of the body size –
has not been found in the Seneca. A space of that dimension could
be made by using two thin spaces.

100%

62
19 Seneca, Quaestiones naturales, sig. l1r

800%

100%

64
A summary of the Aldine norms
Measurement of the primary variables in the design and fabrica-
tion of Griffo’s character sets, such as the ratio of x-height (lower-
case vowel height, in the case of Greek letters) to the mould size /
line-increment of the page, suggests that multiples of the sub-
modular twelfth of the type body size were used in the tooling-up
of the face gauges. My drawing suggests a way in which this could
have been done.
These data, for the character sets discussed here, are given in
the table overleaf (p. 68). Work in progress suggests that the same
interval was used in multiples for gauging the width of punchcut
letters as these fell into dimensionally clustered subsets, groupings
which would have facilitated the supply of blanks for punchcut-
ting in a related set of sizes. This would have minimized effort in
filing down the steel to the required letter widths. Proper space at
each side of a letter depended of course on the skill and experience
of the caster of the types in setting mould registers; the norm then
was for minimal space between letter-face and the sides of a type.
The fine horizontal terminators (serifs), such as those cut by
Griffo for the type used in Bembo’s De Aetna (1495 /6), could func-
tion as spacing bars to guide the caster in making evenly and close-
spaced lowercase characters. When cast as types, then composed
and printed from, a satisfactory image (‘gestalt’) of the word
would be formed. The widths of the metal bodies of the type were
not incrementally determined, as became necessary in the late
nineteenth century with the development of machine composition
driven by compressed air and punched tape. Discussion of the
function of serifs in type design has been plentiful, but it has made
little or no reference to their use as aids to spacing in the setting of
mould registers.
Although the Aldine sub-modular twelfth obviously changed
from one mould/type size to another, my calculations suggest
66
The method used by Griffo to transfer
measurements from combinations of
spacing sorts is a matter of speculation.
This drawing suggests one possibility:
the use of spaces as feeler gauges. (The
spacing sorts need to be shown too large
in relation to the hands.) Needless to
say, both visual and tactile perception
are used in fine metal work of this kind.
From De Aetna (200%).
that the line-increments planned by Aldus after the Lascaris job
could have been compiled from combinations of the duodecimal-
derived subdivisions of the Lascaris em quadrat; that is, from
combinations of the spacing sorts to be found in the Lascaris type
cases. For example, the mould size of the roman cut by Griffo for
Bembo’s De Aetna was the same as that of his second Greek:
namely, eleven twelfths of the Lascaris mould size. Thus, the
body sizing component of the mould used for dimensioning the
De Aetna romans and the second Greek could have been cast or
rubbed to a thickness gauged by reference to two Lascaris ‘thick’
spaces (2 × 4 twelfths) and one Lascaris ‘middle’ space (1 × 3
twelfths), making a total of eleven Lascaris twelfths. Similarly, the
moulds fabricated for the Herodotus Greeks at 8 sub-modular
twelfths of the Lascaris mould, and used to dimension the Greeks,
the cursive Latins and the Arabic numerals required for the com-
position of the octavo classics series at 7.5 Lascaris twelfths, could
have been sized by using two Lascaris ‘thick’ spaces (4 + 4 units),
and a combination of one en space and one ‘very thin’ space (6 +
1.5 units), respectively. This of course does not preclude the use of
punches cut for characters gauged by reference to a mould of one
size being used to strike the matrices destined for use with a larger
mould. For example, the punches cut by Griffo for the Greek capi-
tals that were required for the smaller mould of the octavo classics
type were then used as the capitals of the slightly larger Herodotus
Greek (his third Greek), on which, one can reasonably guess, he
would have been working at much the same time. It may be thought that making a mould
Two sets of relationships are set out overleaf (p. 69): of the to cast spaces for subsequent use as a
scale of reference for smaller or larger
Aldine body sizes to the first mould used to cast the Lascaris types; type bodies is an unnecessarily long-
and of the Aldine quadrats to the traditional name-designated winded process. I do not suggest that
type sizes that were subsequently developed. It seems clear that the spaces in the Lascaris typecases
were specially made to that end, but
the spacing sorts of the Lascaris job could have been used in com-
that they were indeed used for that pur-
bination to gauge an infinitely large set of body sizes, from which pose. My suggestion is that one mould
universe of discourse Aldus had occasion to need but a small sub- served to spawn the system of measure-
set. Insofar as the sub-modular twelfth part of the Lascaris quadrat ment used in the fabrication of other
mould sizes: a kind of typo-genetics.
could have been used as the basic unit of measurement for gaug-
ing the mould sizes of subsequent Aldine type sizes, it seems rea-
sonable to regard this unit (about 0.527 mm) as being the Aldine
typographic point, two centuries before Fournier used such a
term. Not that there was anything special about the Lascaris
mould. Perhaps most printers working before type founding
became divorced from publishing and printing had each his own
master mould?
Just as the combinatorial use of spacing sorts of the first casting
of the Lascaris types could have provided the means for gauging
Type spaces 67
the subsequent mould sizes called for by Aldus, so reference could Dimensional relationship of Aldine em quadrats mm (nearest 0.1)
have been made to them in the tooling-up of face gauges for the
dimensional proportioning and co-ordination of the Lascaris char- 12 6.3 Greek 1
acters themselves. Similarly, the spacing sorts cast from moulds
spawned by the Lascaris mould could have been used for the 11 5.8 Greek 2 /
De Aetna roman
dimensional co-ordination of the faces of the character sets
planned to be cast from such moulds.

8 4.2 Greek 3
Aldine character sets: primary dimensional attributes 7.5 4.0 Greek 4 / italic

Character set Line-increment/ Sub-modular

x-height/ Capital
body size unit (mm)
lowercase height
units (mm)* vowel height (roman style)
units (mm)* units (mm)*
——————————————————————————————————————————
Lascaris, Erotemata (1495), first casting:
Greek 1, cursive 12 (6.3) (0.527) 4 (2.1) 6 + (3.2)†
Latin, roman 12 (6.3) (0.527) 4 (2.1)† 7 (3.7)†

Second casting: 14 (7.4) (0.527) 4 (2.1) 7 (3.7)† 1 [0.525 = 1 Lascaris twelfth]

——————————————————————————————————————————
Greek 2, cursive 12 (5.8) (0.483) 4 (1.9) 8 (3.9)† 0
—————————————————————————————————————————— 12 11 10 9 8 7 6 5 4 3 2 1 0
De Aetna (1495 /6):
Latin, roman 12 (5.8) (0.483) 5 (2.4) 7 (3.4) 12 6.3 Greek 1
ascenders:8 (3.9) 11.5 6.0 (Great Primer, 6.1 mm)
—————————————————————————————————————————— 11 5.8 Greek 2 / De Aetna rom.
10.5 5.5 (2-line Brevier, 5.4 mm)
Herodotus (1502):
10 5.3 —
Greek 3, cursive 12 (4.2) (0.350) 5 (1.8) (2.0)‡
9.5 5.1 —
—————————————————————————————————————————— 9 4.8 (English, 4.8 mm)
Octavo classics series (1501 – 15) and the Seneca (1522): 8.5 4.5 —
Greek 4, cursive 12 (4.0) (0.333) 4 (1.3) 6 (2.0) 8 4.2 Greek 3 (Pica, 4.3 mm)
Latin, cursive 12 (4.0) (0.333) 5 (1.7) 6 (2.0) 7.5 4.0 Greek 4 / italic
—————————————————————————————————————————— 7 3.7 (Small Pica, 3.8 mm)
6.5 3.4 (Long Primer, 3.35 mm)
Wrong font Qu (see Ovid, Metamorphoses, 1502, sig. a5r):
6 3.1 (Bourgeois, 3.0 mm)
12 (3.7) (0.308) 5 (1.5) 6 (1.8)
5.5 2.9 —
—————————————————————————————————————————— 5 2.6 (Brevier, 2.7 mm)
4.5 2.3 (Minion, 2.4 mm)
* To nearest 0.1 mm. Note: the capitals of the second casting 4 2.0 (Nonpareil, 2.15 mm)
† Extant character sets; may not have of the Lascaris types are in the ratio of 3.5
been made by Griffo. 7 : 14 units of the line-increment (mould 3
‡ These capitals are identical in size size). This ratio (1 : 2) was retained for 2.5
and style with Greek and Latin the capitals of the octavo classics series
2
capitals of the octavo series. (i.e. 6 : 12 units of the line-increment).

1 [0.525 = 1 Lascaris twelfth]

68 For old body names, see Carter

12 11 10 9 8 7 6 5 4 3 2 1 0 (1969, p. 127).
Aldine spacing sorts: millimetric equivalents Design analysis 1 Lascaris’s Erotemata and the octavo classics

Lascaris De Aetna Herodotus Octavo classics

hair 1⁄ 0.527 0.483 0.350 0.333 Bibliographers and type historians have focused on letterform It will be recognized that Aldus was not
12
styles, and have measured type size or line-increment in units constrained by a socially standardized
1.5⁄ system of linear measurement, nor by
very thin 12 0.790 0.724 0.525 0.499 (usually millimetres) that have no intrinsic relation to what is an established set of type sizes. Present-
being measured or to the process of its production. Ratios, as day designers are similarly placed in
thin 2⁄ 1.054 0.966 0.700 0.666 they fall on the retina of the reader, have not concerned these respect of our electronic systems of
12
character generation and assembly.
researchers. But concern for ratios is what typographic design
mid 3⁄ 1.581 1.449 1.050 0.999 practice demands. Looking at the Aldine system of measurement
12
below that of the type size can yield tools for the exploration of
thick 4⁄ 2.108 1.932 1.400 1.332 typographic design problems, as distinct from bibliographical
12
problems.
thin + mid 5⁄ 2.635 2.415 1.750 1.665 The three sets of analysis that follow address these design con-
12
cerns. This first analysis entails a comparison between Griffo’s
en 6⁄ 3.162 2.898 2.100 1.998 problem with the Lascaris primer and the character sets required
12
for the diminutive page of the octavo classics series. These books
mid + thick 7⁄ 3.689 3.381 2.450 2.331 became a practical proposition thanks to Griffo’s skill as a punch-
12
cutter of suitably small type faces. Erasmus, who visited the Aldine
thick + thick 8⁄ 4.216 3.864 2.800 2.664 workshops, thought that these italic letters were the neatest in the
12
world.
mid + en 9⁄ 4.743 4.347 3.150 2.997
12

thick + en 10⁄ 5.270 4.830 3.550 3.330 Lascaris, Erotemata

12

thin + mid + en 11 ⁄ 5.797 5.313 3.850 3.663 Lascaris’s primer of 1495 was a two-language book: aimed at the
12
student literate in Latin but new to Greek. The job required Griffo
em 12 ⁄ 6.324 5.796 4.200 3.996 to make a lowercase Greek running-hand on which the student
12
could model his own in the process of learning Greek by writing.
(6.3) (5.8) (4.2) (4.0) This choice was in preference to a formal book-hand such as that
followed by previous printers of Greek. This didactic purpose
seems to have been overlooked by critics of the choice of an infor-
Proctor (1900, pp. 98 – 9) writes of
Mould sizes / line-increments mal hand for the composition of the Greek text. Nor would it be
Griffo’s first Greek: ‘Aldus broke away
true to say, as Proctor did, that the simulation in metal of a freely from the usage of his predecessors, and
Lascaris 0.527 × 12 = 6.324 (6.3) mm written script goes against the grain of the ‘rigid uniformity of produced a type based not upon the
De Aetna 0.527 × 11 = 5.797 (5.8) mm type’. As with stone-cut letters, the form of a punchcut letter is not noble and beautiful older book-hand,
but on the ordinary correspondence
Herodotus 0.527 × 8 = 4.216 (4.2) mm a mechanistic function of tools and materials, as might be said of or business handwriting of his day, in-
Octavo classics 0.527 × 7.5 = 3.953 (4.0) mm the stylus in wax which originally conditioned the formal attrib- volved and contracted to an extreme
utes of Greek script. In the case of intaglio letters made by engrav- degree, but, as writing, not without
merit for its freedom and flowing lines;
ing in copper or steel, the action of the burin is evident. Not so the
and for that very reason eminently un-
use of the file to cut away unwanted metal so as to leave in steel suited for fixing in the rigid uniformity
the shape of a letter in relief, in reverse and of a planned size. of type.’

70 Type spaces 71
The problem of imitating a running-hand in metal is not simply
technical, but may result in the need to store such a large number
of separate pieces of type as would make the type case ergonomi-
cally unmanageable, as the Chinese discovered.
Aside from the didactic function of a Greek running-hand for
Lascaris’s primer, Griffo’s design task can be characterized as that
of being required by his employer to fit a Greek quart into a Latin
pint pot: the ‘pot’ is the dimensions of the mould aperture used to
cast an existing Latin type selected for the composition of the Latin
text of the primer. This type, like the Greek capitals of the Lascaris
which were chosen to accompany Griffo’s new lowercase Greeks,
may well have come from the stock-in-trade of Aldus’s printer,
Andrea Torresani. Torresani had worked for the pragmatist print-
er, Nicholas Jenson, and now owned materials that had belonged
to his late master.
Although the Greek minuscules cut by Griffo for the Lascaris
have been discussed extensively by historians (Proctor, 1900;
Barker, 1992), no mention has been made of the dimensional con-
straints imposed on his task by having to fit the Greeks and their
accompanying diacritics to the field already determined by the
extant Latins – a mould size which turned out to be the snake in
the long grass of Griffo’s task. Not only that. A constraint imposed
on Griffo’s job by the nature of the text of the primer was the need
for the two languages to be composed serially, frequently in the
context of the same phrase or sentence, as well as in parallel (on
facing pages) and independently. This required not only that both
character sets be cast from the same size of mould (that is, have
the same body size), but that both should be anchored to the same
base-line (writing line): an implied horizontal line already posi-
tioned on the vertical axis of the field of play by the extant Latins.
This position provided the learner of Greek script with spatial ref-
erence, by way of the relative positions of the Latin minuscules on
the vertical axis of the three-storeyed writing system, and with
which the student, knowing his Latin letters, was already familiar.
Although the ‘freely written’ Greeks were more wayward in
dimension, form and relative positions in the field than the histori-
cally much more formalized Latins, Griffo chose to make the
lowercase vowel height of the Greeks conform broadly with that of
the extant Latins. These, with respect to the equal length of ascen-
ders and descenders, have the appearance of being centred on the
body of the type. Such a position of the x-height ribbon would
have left precious little room on the body of the type above the
Greek capitals for the accommodation, by whatever means, of dia-
critics, as was traditional. To create the required space above the
72
In a letter to me (4 July 1997) James
Mosley wrote: ‘I agree that the Lascaris
roman is a rather primitive affair. The
caps do have a distinctly Jenson air to
them, as indeed does the lowercase in a
crude way. Mardersteig (‘Aldo Manuzio
e i caratteri di Francesco Griffo da
Bologna’, 1964) hesitates to say that it
can be the work of Griffo, but thinks it
might have been acquired ready-made.
He also thinks that the same roman, or
part of it, was used by Stefano Planck in
1493 for a grammar by Perotti (which
Haebler thinks has a Venetian quality,
as well he might). But against this he
puts Soncino’s assertions that Griffo
made ‘all of Aldus’s types’, and the fact
that Soncino himself used it on a small
scale in 1518.’
Today, when handwriting is very infor-
mal, the writer may not be fully aware
of the position of elements of the letters,
such as descenders, relative to the ‘writ-
ing line’. The lowercase g is one instance
of this, as we found with first year stu-
dents of typography at Stafford College
(c. 1970) following a project designed to
point up conscious awareness of letter-
form details. In some cases, the lack of
awareness of the three-storey structure
of the writing system was astonishing.
This may have been a consequence of
the fact that teachers of young children
in the uk have not themselves been
taught how to teach the rudiments of
handwriting by personal example.
Greek capitals for the positioning of diacritics (sometimes double-
deckers), the writing line/base-line, which was common to both
languages, was set very low in the field as defined by the mould
aperture. As first noticed by Proctor (1900, p. 100), this required While the term ‘mould aperture’ is
the descenders of both languages to be kerned downwards. technologically specific, the equivalent
term, ‘character field’, is applicable to
Although this may be counter-intuitive, the alternative strategy both metal and electronic systems and
of up-kerning the diacritics to rest on the ‘beard’ of the previous provides the connection between the
line would have required many more ‘up-kerns’ than ‘down-kerns’, two. See the teaching aid made by May
and Burgess (p. 36).
and was probably pre-empted by a decision to make them inter-
changeable with host characters by lateral and counterwise kern-
ing. Like lampreys, a surfeit of kerning is Not a Good Thing. This
is especially true of down-kerning where damage to the delicate
overhanging parts can be caused not simply by pressure from the
platen of the printing press, but also by having to slide the types
being set under the kerns of the previous line; the lines being set –
as read – from left to right, but upside down (see figure 26).
This obstruction could have damaged the kerns in the process of
typesetting, and may have been the reason for the decision to
recast the Lascaris types using a mould aperture larger by two sub-
modular twelfths of the initial mould. The increase in body size by
that amount obviated the need for down-kerning (see figure 29).
A side-consequence of a larger mould size was a change in the
ratio of capital letter height to line-increment, from 7 : 12 to 7 : 14.
This mutation – the new ratio (1 : 2) – became the norm used by
Griffo for the capital height to line-increment ratio of the com-
bined set of Greek and Latin capitals of the octavo classics series.
A further side-consequence of the standardization of capital
heights at half the line-increment/type size, was the opportunity
that this gave for the letter and word-spacing of all-capital set-
tings, and for this to be done with no danger of contravening the
logic that demands more space to be seen between lines than can
be seen between words. Although Aldus retained the generous In his homage to Aldus, Anthony
line-spacing created automatically by the standard established Froshaug also used {capital height × 2}2
as his modulus in page design. (See
for the capital height to line-increment ratio, he seems to have pp. 132 – 4.)
abandoned the traditional practice of housing diacritics above
the capitals of the Greek.
The following set of illustrations (figures 20– 27) relates to
Lascaris’s Erotemata of 1495. Having to fit the lowercase Greeks
into the character field of the extant Latins, so that the two lan-
guages could be composed sequentially, caused problems which
the following analysis attempts to make clear, and which later led
to better solutions.
Type spaces 73
20 Lascaris, Erotemata, first casting, opening pages

50%

The first line is set against the flange of the woodcut; accents are
kerned laterally to rest above the capitals. Note that the composi- 400%
tor would see the matter upside down, and from bottom to top.

100%
Imprints of the top edges of the type low base-line of the Lascaris types, with The extant type chosen for the Latin
bodies shown here and in the illustra- the consequent down-kerning of both text. Note the damaged ‘p’.
tions opposite are evidence of the very Greek and Latin descenders.
21 Lascaris, Erotemata, first casting, from sig. g4

300%
100%

50%

100%

400%
22 Lascaris, Erotemata, first casting, sig. h3: the first Greek
23 Lascaris, Erotemata, first casting 24 Lascaris, Erotemata, first casting, sig. q4

100%

100%
Note very close word-spacing; also the broken ‘q’.
400%

400%

At first sight it might appear that the x-height of the Lascaris

roman is centred on the body, as it probably was in its original
state before its matrices were used for this job.

Greek and Latin share the base-line and have a common x-height.
Note (asterisk) the clash between kerned descender and the dia-
critical mark housed in the space created above the Greek capital Type spaces 79
by the low base-line.
25 Lascaris, Erotemata, first casting: first Greek 26 Lascaris, Erotemata, first casting: first Greek

100% 100%

Source page is shown in figure 20.

Analysis of structure A compositor’s eye view of the type as set: upside down, and from
bottom to top; with a conjectural view of kernings.
400%
27 Lascaris, Erotemata, first casting, sig. a2r
On this page Aldus sets out for the stu-
dent a few of the alternative characters
to be encountered in the text. Many
were made, to facilitate lateral kerning
and the simulation of a running hand
on which students could base their own.
The total number of pieces needed to
compose this first Greek was once esti-
mated at 1,400 (by Wolfgang Fugger in
1553, quoted by Harry Carter [Fournier,
1764, vol. 3, p. 228, note 1]). This is four
times the number in his fourth Greek:
variety reduction then led to a more
manageable character set and a clearer
text.
The octavo classics project
With Lascaris’s Erotemata, Griffo’s task had been to squeeze a Between 1501 and the death of Aldus in
Greek cursive into a field determined by the mould size of an exist- 1515, a total of 43 titles were published
in the octavo format. Of these, 8 were in
ing set of Latin types. The new smaller page selected for the pro- Greek, 33 in Latin, and 2 in Italian.
jected classics series demanded that another mould size be found,
giving a maximum permissible depth of thirty-two lines of text.
This depth left enough margin space to withstand the variations in
trim that could easily occur when binding was not under the con-
trol of the Aldine workshop. Many copies of an edition would have
left the workshop unbound or just in printed-sheet form, for sale
in that form and thus for finishing elsewhere in Europe.
The choice of thirty-two lines maximum for the octavo page
would have provided a first approximation of the required mould
size/type size: a dimension to be finalized by reference to some
multiple of the sub-modular twelfth of the Lascaris mould, as dis-
cussed above. Factorially, thirty was a ‘good number’ if arbitrary
damage to the stanzaic structure of classic verse was to be mini-
mized at page turn-over points. The two extra lines (1 and 32) were
normally suppressed, except when needed for such things as run-
ning heads, page numbering when used, catchword and signature
lines, and for the housing of the unpredictable over-run lines that
can occur in the making-up of pages from galley-stored prose
matter, mixed verse and prose, and different varieties of listed
information in single- and two-column form. Textual variants of
all these kinds are to be found in the typographic composition of
books of the octavo classics series. The possible occurrence of
combinations of these kinds had to be accounted for in the context
of dimensional decision-making in advance of production.
The maximum permissible width of the type area in the metal, See pp. 101 – 12 for reconstruction of the
but not always as seen on the printed page, was set at 18 ems of the control system used in the page assem-
bly of the octavo classics series.
type size. This dimension seems to have been determined by the
two-column structure of the index, the lines being composed ini-
tially in a single column of 9 ems measure, and subsequently
arranged as two parallel columns at the page make-up stage of the
job. The intercolumn space was created by starting each entry of
the 9 em column with a ‘middle’ (3 unit) space. As a consequence,
the overall measure in the metal of the two-column page had per-
force to be an even number of ems. A measure of say 20 ems would
have left little margin for possible variations in trim at the fore-
edge of the octavo page. On the other hand, a measure of 16 ems
might have been seen as a profligate use of expensive paper, mak-
ing the book more expensive than it might otherwise have been.
So, an overall measure of 18 ems of the type size had it. All other
Type spaces 83
28 Ovid, Metamorphoses, 1502
The 32-line page of the octavo classics
series shown at real size. Note the vary-
ing dimensions of the intervals follow-
ing the stand-off markers, suggesting a
fine-grained spacing system. Note too
that one line has exceeded the predeter-
mined measure and has been turned
over to be set at the end of the subse-
quent line. Of the 12,000 lines of the
poem, I could spot only two that had
had to be broken arbitrarily in this way.
Lines 1 and 32 were reserved for
extra-textual information or were
suppressed: that is, set as blank lines.
Here line 1 carries the title, line 32
is suppressed.
100%
84
textual components seem to have been set to this measure.
Variations in the overall width of the printed image of the text
were achieved by differing amounts of space being set at the start
of the lines (see figure 44). In this way, only two fixed-width com-
posing sticks would have been needed, one of 9 and the other of 18
ems of the type size. Constraints of this kind lead one to question
the notion that type areas were determined by some ideal geomet-
rical division of the page area. The evidence presented here sug-
gests that the decision was a matter of weighing up the pragmatic
and syntactic needs of the work to be done.
It has sometimes been said that Aldus chose the italic style for Morison (1926, p. 26) says of Griffo’s
the Latin text of the octavo series because italic forms are laterally italic and its use in the composition of
the octavo classics series: ‘Its chief merit
compressed, and were thus appropriate for the narrow page of the is that of making up extremely econom-
series. In fact, Griffo’s italic is not particularly compressed. It is the ically, a supreme quality in such a series
x-height of his letters relative to the type body size which makes as that for which Aldus employed it.’
Against this, Carter (1969, p. 74) writes:
the width of the characters economical on the horizontal of the
‘If Aldus hoped, as it is commonly said
page. The nicety in choosing only 5 sub-modular twelfths of the that he did, but he never said, that cur-
type size for gauging the x-height of the italic is illustrated by the sive letterforms would save space, he
fact that of the twelve-thousand lines of verse of the Ovid, I could must have been disappointed by the
results: a Roman type on the same body
spot only two which had to be broken arbitrarily at line endings gets in just as much.’
(figure 28). The small x-height, in relation to the line-increment
of the work, was a consequence of the reinstatement by humanists
of the dignity and letter-defining function of the ascenders and
descenders of the Latin minuscules, after their truncation in
medieval formal writing. My feeling is that this was probably done
originally to get in more lines to the page without reducing the
retinal image size of the text, as gauged by reference to the depth
of the x-height of the work. I like to think that Aldus chose the It may be recalled that Monotype Times
italic for the Latin and vernacular text of the octavo classics series, Roman had truncated extenders (and
thus a large x-height). This was done to
not simply because it was the mark of new learning, but also as an get in as many lines as possible, while
act of homage to its inventor, Niccolò Niccoli, who was a conduit retaining a large retinal size in relation
of classical manuscripts into Italy: manuscripts of the kind which to the line-to-line dimension, in the
composition of The Times newspaper.
Aldus, a century later, was to mass produce with due elegance by
In that case, the characters give the
way of the new technology of printing. appearance of being condensed on the
horizontal axis of the field, as did the
For the Greeks of the octavo series, Griffo chose a lowercase vowel late medieval ‘gothic’ letter.
height of only 4 sub-modular twelfths of the type body size which,
when perched on the same writing line as the 5-unit, body-centred
Latin cursives, provided 4.5 units on the body above the mean line
of the Greek minuscules for the accommodation of diacritics
(figure 34). This left 3.5 units on the body below the common base-
line of both languages for the deployment of descenders, without
Type spaces 85
recourse to down-kerning. For the Greeks of this series, the
accents and breathings may have been made interchangeable with
host characters at the matrix-striking stage of production by piggy-
back riding, both being struck together to the same depth, as is
necessary if an evenness of printed image is to be maintained.
Some evidence exists to indicate that a type size smaller than 7.5
Lascaris twelfths may have been contemplated for the octavo clas-
sics project. This evidence takes the form of a single-bodied ‘Qu’
smaller than the norm which appears at sig. a5r of the Ovid of 1502
that I have examined (figure 44). If the gauging of the Q and the u
of this ‘wrong font’ follows the system of sizing discussed here,
then that diminutive pair could have belonged to a trial run using
a mould body of only 7 Lascaris twelfths (see table, p. 68). Perhaps
such a small face for the italic, and an even smaller face for the
Greek, proved too difficult to cut, the faces being among the first
movable types for printing to be made so small. Be that as it may,
in the event, the 7.5 mould was used for gauging the Latins, Greeks
and Arabic numerals required for the projected series of pocket-
sized books.
As described above (p. 67), the capitals used to accompany
the lowercase of the third Greek seem to have originated in the
punches cut for the capitals of the fourth Greek, as sized by refer-
ence to the 7.5 mould, not to the slightly larger mould (8 Lascaris
twelfths) used for the lowercase Herodotus Greeks that were
made for printing Greek-only texts in folio formats.
In view of the slight difference in size between the 7.5 and the
8.0 mould – a difference of 0.2 mm – the question arises as to why
the lowercase punches of the third Greek, suitably recut where
necessary, were not also used for the octavo classics project.
Although occasional use was made of them in that context (for
example, the Juvenal and Persius of 1501), the mean height of the
rotund lowercase of the third Greek, gauged at 5 sub-modular
twelfths of the 8.0 mould, may have made the face too large – too
greedy of lateral space – for the narrow page of the octavo series.
For that page, a smaller Greek – the fourth – was necessary; this
had a lowercase vowel height of only 4 units of the 7.5 mould
(figure 34). It satisfied the constraints on line length and page
depth imposed by the octavo format. For this type a greatly sim-
plified character set was developed, with many fewer sorts than
Griffo’s earlier Greek cursives had used, and this established a
standard around which the Greeks of the next few centuries were
to rally.
86
Barker (1992) argues that, in the case of
all Griffo’s Greeks, the diacritics were
separately cast using a very small mould
and composed in a separate line above
the mean-line of the minuscules, some-
times involving as many as three
moulds for a single line of printed
Greek. How the compositor could set
such a complex structure in the compos-
ing stick while moving left-to-right,
upside-down, and with the characters
cast in reverse, he does not examine. He
also posits the view that a standard unit
of measurement used by the early print-
er for mould sizing may have been that
used by fine metal workers of the day. ‘I
wish’ he wrote (p. 79) ‘that I had seen a
contemporary ruler that would confirm
(or not) this conjecture’. The unit sug-
gested by Barker as the metal workers’
standard is the ‘hair’ space of the
Lascaris mould.
The second Qu is smaller than the first:
a ‘wrong font’. See figure 44 for further
discussion.
Character sets having subtly differing
sorts of the same character would pre-
sent our electronic storage systems with
few difficulties compared with the com-
positor’s type case and its use in hand
composition.
As seen through the distorting lens of a later formalism, it might be
said that the capitals of the octavo series are too small. Or is it that
the x-height of the italic is too large? Either way, such comment
ignores Griffo’s design problem: that of sizing a set of capitals to
serve both Greeks and Latins when used in series and separately,
and when the Greek minuscules were required to be smaller than
the Latin, in relation to the line-increment of the job, to allow
room above the mean line of the Greeks for the housing of diacrit-
ics. In making the capitals of both languages the same in size and
styling, Griffo overcame a problem which is clearly evident in the
printing of the Lascaris primer of five years earlier. There, the very
slight differences observable between the two sets of extant capi-
tals caused them to become mixed in use. One can imagine that
the compositor was unable to distinguish clearly between the two
sets when seen in reverse and stained with ink, while distributing
types to their respective cases, to be ready for reuse. As many
roman and Greek capitals have the same form, making the two
sets conform in size and detailing would help to overcome this
problem and minimize the work of punchcutting and casting
when it came to the similar two-language task of the octavo clas-
sics series.
All-in-all, Griffo’s work for the company during the last decade of
the fifteenth century illustrates the design process, as discussed by
Gombrich (1984) after Popper, as essentially one of trial and error
correction. Although the work that Griffo did at the call of Aldus
set the stage for the future development of Greeks, romanized
Latins, and the italics – all of which today we know so well – the
unified system of dimensional control used in-house by the firm,
and applied at all levels of order, including character image sizing,
has still to be fully realized. I think there is a sense in which the
Aldine control system may be seen as the precursor of the poten-
tial which is present at the heart of the means we now use for
designing typefaces and setting text. In this regard we should
remember that the human visual system is the same now as it was
when Griffo was working near the limits of vision, using a sub-
modular unit of measurement discernible with little if any optical
assistance. Equally, the intrinsic structure of human language is Noam Chomsky’s work provides the
the same, irrespective of the typographic conventions we develop notion that human languages are struc-
turally similar.
to set its intrinsic order clear to view.
Type spaces 87
Design analysis 2 Aldine character sets: relative proportions

The following notes show the relative proportions of the major

dimensional attributes of Griffo’s character sets: within a set,
between sets of differing type body size, and between Greek and
Latin sets of the same and different type sizes.

Type spaces 89
29 First Greek, first casting (mould size: approx. 6.3 mm) 30 First Greek, second casting (mould size: approx. 7.4 mm)

100% 100%

400%

400%

x-height / line-increment ratio – 4 : 12 units (1 : 3) x-height / line-increment ratio – 4 : 14 units (2 : 7)

capital-height / line-increment ratio – 7 : 12 units capital-height / line-increment ratio – 7 : 14 units (1 : 2)
x-height / inter-x-height ratio – 4 : 8 units (1 : 2) x-height / inter-x-height ratio – 4 : 10 units (2 : 5)
31 Second Greek (mould size: approx. 5.8 mm) 32 De Aetna roman (mould size: approx. 5.8 mm)

100% 100%

400%

The ‘i’ in ‘interdum’ is inverted; its horizontal alignment indicates

that the x-height is centred on the body.

400% 100% Although this

character set and
the second Greek
have the same body
size, they would not
align if set sequen-
tially. Note the
rising word-space
in the enlarged
segment.

400%

x-height / line-increment ratio – 4 : 12 units (1 : 3)

capital-height / line-increment ratio – 8 : 12 units (2 : 3)
x-height / inter-x-height ratio – 4 : 8 units (1 : 2)
x-height / line-increment ratio – 5 : 12 units
capital-height / line-increment ratio – 7 : 12 units
x-height / inter-x-height ratio – 5 : 7 units
33 Third Greek (mould size: approx. 4.2 mm) 34 Fourth Greek (mould size: approx. 4.0 mm)

100% 100%

400%

The Greek–roman
set of capitals cut
for the octavo clas-
sics series was also
used as the capitals
for the third Greek,
cast on a slightly
larger body.

Sophocles, 1502
(bl.687.b.2)

400%

x-height / line-increment ratio – 6 : 12 units (1 : 2)
capital-height / line-increment ratio – 6 : 12 units (1 : 2)
x-height / inter-x-height ratio – 6 : 6 units (1 : 1)
The capitals used with this Greek –
the lowercase of which was made for
use on a folio page – are identical with
those sized for use with the 4 mm of the
octavo classics mould.

x-height / line-increment ratio – 4 : 12 units (1 : 3)

capital-height / line-increment ratio – 6 : 12 units (1 : 2)
94 x-height / inter-x-height ratio – 4 : 8 units (1 : 2)
35 Latin cursive (mould size: approx. 4.0 mm)
100%
Ovid, Metamorphoses, sig. q1v
400%
x-height / line-increment ratio – 5 : 12 units
capital-height / line-increment ratio – 6 : 12 units (1 : 2)
x-height / inter-x-height ratio – 5 : 7 units
36 Latin cursive and fourth Greek (mould size: approx. 4.0 mm)
300%
Having to fit a Greek quart into a Latin pint pot proved to be the
main stumbling-block in the Lascaris job of 1495. Griffo’s problem
of combining Greeks and Latins sequentially was beautifully
resolved some five years later when it came to making and co-
ordinating the Greeks and Latins needed for the composition of
the thirty-two line page of the octavo classics series, the last of the
jobs he did for Aldus.
100%
At this stage in the development of
Greek type, the letterforms had not yet
been ‘colonized’ by the Latins, and still
carried the mark of stylus in wax.
Ovid, Metamorphoses, sig. a2v
100%

37 Ovid, Metamorphoses, sig. d1v–d2

400%

Spaces cast in multiples of one twelfth of the em quadrat of the

type’s body size provided a scale of values for determining not only
such primary variables as x-height and capital-height, but all
intervals less than the modulus (the em space) – on both axes of
the page.
38 Schematic representation of dimensional relationships
shown at the same sub-modular unit size
First Greek, first casting
Quadrat: 6.3 × 6.3 mm
Second Greek
Quadrat: 5.8 × 5.8 mm
Third Greek
Quadrat: 4.2 × 4.2 mm
(capital-height as fourth Greek)
First Greek, second casting
Quadrat: 7.4 × 7.4 mm
De Aetna roman
Quadrat: 5.8 × 5.8 mm
Fourth Greek & italic
Quadrat: 4.0 × 4.0 mm
x: lowercase vowel-height or x-height
c: capital-height
a: ascender / descender range
d: diacritic range
Design analysis 3 Typographic reference grids
The following illustrations examine dimensional relationships in
the composition of textual components of differing kinds as found
in the page make-up of the Ovid of 1502 and the annotated Seneca
of 1522. The analysis is based on the assumption that the adjust-
able composing stick was not then in use, but that the metal was
set to fixed measures determined by analysis of the job to be done
I do not wish to suggest that typographic reference grids as shown In the uk, the typographer Maurice
here were used by Aldus to communicate his typographic design Goldring and his partner Angela
Hackelsberger were leading exponents
requirements to his shop-floor workers. I show them as a form of of the reference grid: an aspect of a
reverse engineering to support the main thesis of this essay into specification system not to be confused
Aldine typographic norms. On the other hand, the dimensional with the formalistic ‘grid’ of the Swiss
typographers of the mid-twentieth
control system as used in the manuscript tradition was left scribed
century.
on the page for all to see. As an artefact, such a control system was
built into the tools and materials of the new technology and would
have been well understood by the early printer and by scriveners
retrained as compositors, whatever may later have been forgotten
or never known by printing trade workers. Only in the twentieth
century, when the typographer became an architect of the printed
document, was the reference grid developed as a distinct compo-
nent of an overall specification system for print production. Up to
that time it had existed, if at all, in a rudimentary form known to
compositors as a case-room layout or a layout artist’s sketch.
Type spaces 101
39 Ovid, Metamorphoses, sig. a3v–a4

Note: this page was not bound in square

when the book was trimmed

← line 1

← hyphen missing at line end, causing

initial ‘A’ to be misplaced

100%

The single- and double-column reference grids shown overleaf,
and worked out retrospectively from the double-page spread
shown here, specify the major dimensional variables in the typo-
graphical construction of the pages of the octavo classics series
(1501 – 15). How Aldus co-ordinated the work of his printing staff
must remain speculative. However, as little appears to have been
left to chance in the detailing of the work, it seems reasonable to
suppose a carefully worked-out set of house rules for use at both
macro and micro levels of order; rules which were dependent on
the use of a unified system of modular and sub-modular control.
Type spaces 103
40 Octavo classics series: reference grids

Taken from figure 39, these single- and

double-column reference grids indicate
the primary dimensional variables gov-
erning the design of the pages of the
octavo classics series.

104 Type spaces 105

41 Ovid, Metamorphoses

Use of single-column reference grid

showing a mixture of prose and verse.
Note the use of stand-off capitals and of
indention beyond the left-ranging verti-
cal axis of the lowercase. Lines 1 and 32
are suppressed in this instance. Note
mid- and high-level use of the ‘full
point’.

100%

106 Type spaces 107

42 Octavo classics: page make-up 43 Typesetting procedure
using a 9 em composing stick (conjectural)

Single column: set to 18 ems of the type 1 Set one em-quadrat, followed by one
size. mid-space.
Double column: set initially to 9 ems of 2 Set and justify the line, excluding the
type size. stand-off capital.
3 Take out the em-quadrat; move mid-
space to the left.
4 Set the stand-off capital after the mid-
space.
5 Make the line tight by setting spaces
after the stand-off capital.
If no stand-off cap, do 1 and 2 only.

It should not be deduced that the set-

widths of the type bodies are a precise
number of sub-modular units. The
tightening-up of the matter in the chase
would take up any slight slack.

Ovid, Metamorphoses (100%)

Ovid, Metamorphoses (100%)

Detail (200%)
Detail (200%)
44 Typesetting procedure 45 Dimensional attributes of textual components
using an 18 em composing stick (conjectural)

1 Set two em-quadrats and one mid- The maximum permissible measure in the design of the octavo
space. series was set at 18 ems of the type body size (see p. 83 for further
2 Set and justify the line, excluding the
stand-off capital.
discussion). All textual components were set to conform to this
3 Take out the second em-quadrat; standard, as illustrated below and overleaf.
move mid-space to the left.
4 Set the stand-off capital after the mid-
space.
5 Make the line tight by setting spaces
after the stand-off capital.
If no stand-off capital, do 1 and 2 only. 1 Continuous prose matter, visually
‘justified’. Note close word-spacing and
letter-spaced caps.

2 Two-column index with vertically listed

information structured alphabetically;
visually justified.

3 Horizontally listed information with

a space of 1 en after each entry, to aid
visual discrimination.

Detail (200%)

The smaller Qu suggests that a smaller

mould size may have been considered
for the octavo classics series. 4 Mixed prose and vertically listed in-
formation structured alphabetically
Below (800%): in single-column form, with numbers
standard Qu (left) ranged right. Stand-off caps are used
wrong font Qu (right) Ovid, Metamorphoses (100%) to mark the start of an entry, with run-
ons indented 1 em (12 units).

Ovid, Metamorphoses (100%)

46 Dimensional attributes, continued in ‘galley proof’ form 47 All-capital headings

5 Mixed prose and verse. A stand-off

initial marks the start of both prose and
verse parts. Note that sub-headings, set
in lowercase, are not centred but ranged
left to a secondary vertical axis. In
Aldine typography, line-spacing for the
functional grouping of parts on the ver-
tical axis of the page is used sparingly,
perhaps to save paper – in the sixteenth
century, the most expensive component
of the manufacturing process. Note also
the use of high and mid positions of the
‘full point’. (On the development of con-
ventions in punctuation, see Parkes,
1982.) Generally there are few typo-
graphic conventions used today that
were not subject to experiment by
Aldus.

Note: This ‘galley proof’ has been
simulated by pasting together copies
from successive pages. In order to make
it fit here, immediately above the last
line (with ‘g’) three lines of text have
been cut.
The ad-hoc spacing of the characters and character groups of top: Lascaris, 1495
the all-capital headings of the Lascaris primer of 1495 can here be middle: Ovid, 1502
bottom: Seneca, 1522
compared with the all-capital headings of the Ovid of 1502 and the
Seneca of 1522. In the case of the Ovid, the capital-height to line-
increment ratio is 6 : 12 units (1 : 2), as against the 7 : 12 units of
the Lascaris primer (first casting). The extra line-space seen in
the all-capital features of the Seneca is created by setting en spaces
(6 unit spaces) sideways-on between the otherwise ‘set-solid’
lines. All examples are shown at real size.
In the design of the character sets intended for the octavo For an extended discussion of the
classics series, gauging the capital height at half the line-to-line- contribution of Aldus to the develop-
ment of ‘small caps’ as a functionally
increment (half the mould size) provided an opportunity to group distinct sub-set of the set ‘capitals’, see
the component parts of the language rationally, compared with Smith (1993).
the all-capital massing of the ill-spaced character groups of the
Lascaris heading.

Ovid, Metamorphoses (100%)

Type spaces 113

48 Seneca, Quaestiones naturales
50 %
Seneca, Quaestiones naturales,
sig. p1v–p2, showing unfolded corner.
Left: watermark (actual size)
For watermarked monogram,
see figure 50.
The Seneca of 1522 is composed in the Latin and Greek cursives ‘Ce livre est depuis long-temps regardé
and the Arabic numerals fabricated around 1500 by Griffo for comme un des plus rares de ceux qu’ont
imprimés les Aldes.’ (Renouard, 1834).
Aldus’s octavo classics project, which ran from 1501 to 1515, when
the publisher died. Although octavo in format, the Seneca is print-
ed eight pages to view on a sheet larger than that of the octavo
classic series; hence the term ‘large octavo’ in describing the size of
the Seneca.
This copy of the Seneca appears to have been rebound and
retrimmed to its present page size of 197 × 130 mm. An accidental-
ly folded-over corner (sig. p2, also carrying the page number ‘116’)
has escaped the plough at the second binding, suggesting that the
page size after the first binding was some 202 × 137 mm. The paper
is laid, watermarked and with a monogram, suggesting that the
firm was having paper specially made, perhaps as an anti-forgery
device.
Single- and three-column structures are used in the composition
of the work, the maximum permissible print area being 22 × 41
ems of the type body size (1 em = approximately 4 mm), lines 1
and 41 being reserved for extra-textual material.
Analysis shows that composing sticks of three different meas-
ures were used in the construction of the single- and the three-col-
umn structures; namely, 22 ems, 7 1⁄6 ems and 7 1⁄3 ems (see figures
49 and 51). The inter-column spaces were created by setting a
space of one en (6 units) at the start of each entry, run-ons being
indented one em (12 units). The three-column list of errors (col-
umn width: 7 1⁄6 ems ) fell short of the 22 ems overall measure by
one en space (6 units). A space of this width was added to each
line of the first column at the page make-up stage of the work, thus
bringing the overall measure to 22 ems. In the case of the three-
column index, set at 7 1⁄3 ems per column, no extra space needed to
be added to the first column to bring the three-columns to the
overall measure of 22 ems. It is clear from this that the 2 unit ‘thin’
space (1⁄6 of the em) is an integral element in the structure of the
Aldine norms, as distinct from the Anglo-American ‘thin’ space
(1⁄5 of the em).
Type spaces 115
49 Seneca, Quaestiones naturales: list of errors (sig. q7v–q8)

100 %
50 Seneca, Quaestiones naturales: watermark monogram 51 Seneca, Quaestiones naturales: first page of index (sig. r1r)
(sig. e5r)

100 %
52 Seneca, Quaestiones naturales: index (continued) and
colophon (sig. r5v–r6)

100 %
The mould 53 Flemish mould

I am grateful to my friend and colleague Alan May for pointing out

to me the construction of the Flemish mould as illustrated and
mentioned, almost in passing, by Fournier. In a footnote to his
translation of Fournier (1995, vol. 3, p. 197, note 1), Harry Carter
says of the Flemish mould: ‘The chief differences between the
French mould and the Dutch are that the latter has no bottom-
plates, the carriages serving as the basis of the structure; that the
Dutch can be justified for body, i.e. the bodies are made easily
detachable from the carriages so that by inserting others the same
mould can be used for several sizes of type, the jet being made
adjustable to suit the cavity ...’ Carter goes on (p. 198): ‘Henze’s
Handbuch der Schriftgiessery, 1844, p. 119, says that the same
mould is made to serve for nine sizes from pearl to pica by the
insertion of different bodies.’ Given that North-European printers
were among the first to set up shop in Italy, it seems possible that
the Aldine mould had interchangeable body sizing components of
this kind. Fournier himself says (1995, vol. 3, p. 197) of the kind
used in ‘Holland, Flanders and elsewhere’ (my italics) that ‘nearly
all their parts are of cast copper’, including it may be presumed,
the interchangeable body part. In his note to Carter’s text, James
Mosley makes a qualification to the translation: this mould ‘was
made largely of brass’ (1995, vol. 3, p. 360).

Mould used in Germany, Holland, and
elsewhere
1. Bottom-half.
2. Top-half.
3. One half seen from below to show
the way in which the jet and the stool
are fastened to the carriage.
4. Jet. The two screws on either side
pass through the two slots and fit the
threaded holes to be seen in the car-
riage beneath.
5. Carriage, hollowed at the back to
lessen its weight.
6. Body, fastened to the carriage by a
screw whose head is countersunk in the
body, and whose other end fits a thread
in the carriage.
7. Screw which fastens the body.
8. Nick, laid on the carriage and held (Image taken from Fournier, 1995,
in position simply by the body, which vol. 1, plates at the rear of the book; the
has a cavity to receive the round head of text here is Harry Carter’s translation of
the nick. Fournier in this edition: vol. 3, p. 305;
9. Wing. It acts instead of the male see also Mosley’s note at vol. 3,
gauge. The two halves are kept on a pp. 384 – 5.)
level with one another by the wings and
the registers. The screw shown above it
fastens it to the carriage.
10. Register, fastened to the carriage
by the screw shown underneath.
11. Stool, fastened to the outer side of
the carriage by the screw shown beside
it.
12. Cast letters as they appear when
taken from the mould. They are seen
from three different angles, and with
and without the cast jet or ‘break’.

122 Type spaces 123

Coda
In these notes, I have been concerned to examine my hunch that
an in-house system of dimensional control existed in typographic
design practice before typefounding became divorced from the
process of planning, composition, and reproduction by printing.
I have argued that, in the case of Aldine publications, the printer’s
laterally adjustable mould was the instrument of his system of
measurement at both modular and sub-modular levels of organ-
ization, and on both axes of the page. That is, the control system
which makes the reading process possible was an attribute of
the production system, not imported or superimposed. In this
regard, I believe it was not unlike the width of the scribe’s square-
ended pen which not only determined stroke-width but was prob-
ably used to pace-out the distance between writing lines (line-
increments) and the dimensions of such a basic parameter in the
construction of the alphabet as the x-height. The printer may have
taken his cue, with suitable modifications, from the practice of
writing – as he did with so much else.
Unlike Moxon and Fournier, whose basic unit of sub-modular
control appears to have been disjoint relative to the duodecimal
division of the em quadrat, Aldus used a unified system which,
among other things, enabled him to know the actual size of the
printed images of types relative to the line-increment of a job.
Unless character image sizes can be specified by designers in visu-
ally discernible units of a unified system of measurement, such
basic design parameters as the ratio of x-height and capital-height
to line-increment cannot be quantified rationally, or made subject
to algorithmic construction.
Aldus was a grammarian who came late to printing. Printers
interested in the dimensions of spacing material have focused on
the technical problems of the justification of three-dimensional
metal types. Aldus’s difference from such printers, and his distinc-
tion, was his concern to find a match between the linguistic and
the dimensional structures of the matter to be composed typo-
graphically: a form-generating equation first designated ‘isomor-
phic’ by Anthony Froshaug and which subsumes the technical. The
anomaly of typographic planners not being able to know the actu-
al image size of characters seen on paper or screen is of long stand-
ing. It is perpetuated still by engineers who seem not to be sensible
of the relationship between measurement and the use and struc-
ture of the language in its visible mode. It hasn’t always been so.
124
I was taught this method by Mr Blunt at
my elementary school in the early
1930s.
Supplement 1 Positioning for meaning
An aspect of Aldine typography shown in these pages is the use of
secondary ranging axes for the positioning and vertical alignment
of the starting point of sub-headings of common structural value.
Such ‘asymmetry’ in the positioning of textual components – as
distinct from the use of bilateral symmetry (‘centring’) – could be
considered to be a characteristic of the ‘modern’ typography of the
twentieth century. In fact, Aldus used both bilateral and displace-
ment symmetry in his design of pages. In this regard, he seems
to have had no ideological worries: as have sometimes bedevilled
the work of typographers of our own day, in their search for a the-
oretical foundation for practice rooted in the formal attributes of
abstract art, as distinct from the search for a one-to-one corres-
pondence between the linguistic order of a text and its visual/
dimensional realization.
It is noticeable that before the advent of movable types and
spaces, which lent themselves to the easy, lateral movement of
whole words and phrases in the hod of the composing stick,
‘centring’ is not commonly found in manuscript practice, being
difficult and time-consuming to do well. In the hand-written
mode of character assembly, ranging left from a predetermined
vertical axis is perhaps the more natural way to proceed; princip-
ally because changing word spaces arbitrarily, once they have
been made, is not an option, as it is in the typographic mode of
character assembly.
Type spaces 125
54 Martial, 1501 55 Martial, 1501

↓ secondary ranging axis ↓ secondary ranging axis

100 % 100 %

← one line space

← one line space

← space for five-line initial

← subheading

← rising word space

← rising word space

← subheading

(rul) (rul)

32-line page (maximum permissible). 31-line page (line 32 suppressed).

Note: secondary ranging axis for positioning the start of subhead- Major headings, as an ‘over-all’ feature, are more or less ‘centred’,
ings, an asymmetrical feature befitting the ‘left-ranged and open- as logic demands. ‘Symmetry versus asymmetry’ was not an ideo-
ended’ character of printed verse. logical issue, as it was to become in the ‘modernist’ period of the
twentieth century.

126 Type spaces 127

56 Martial, 1501 57 Pontano, 1505

↓ secondary ranging axis

100 % just under 100 %

← one line space

← half-line space

← half-line space
(made by setting en quadrats
sideways)
← half-line space

← half-line space

← rising word space

← rising word space

(rul) (rul)

31 lines (line 32 suppressed). 32 lines. Wood-cut initials.

Line spaces, as a means of grouping, are used sparingly in Aldine
typography; presumably to economize on the use of paper.

128 Type spaces 129

58 Pliny, 1509 59 Sophocles, 1502

100 % 100 %

← half-line space

← half-line space
(made by setting en quadrats
sideways)

← rising word space

(rul) (bl.687.b.2)

Note: punctuation marks are centred in the space, and clearly

formed.
31 lines (line 32 suppressed).
Aldus’s fourth Greek. Note how few capitals are used.

130 Type spaces 131

Supplement 2 Anthony Froshaug’s homage to Aldus

This work was designed & printed, presumably in just a few copies,
by Anthony Froshaug in 1950: his number for the job (‘50088’) at
the bottom tells us that it was the eighty-eighth job of that year.
Froshaug once, in 1972, reproduced this piece and wrote his own
caption, as follows: ‘Roman caps with italic lowercase, in tribute to
Aldus Manutius, 1501. All major problems of orthography, punctu-
ation and layout of orthodox texts were solved in the first 100
years of typography. Note escalated indents from left and right
and punctuation spaced by phrase.’ (Kinross, 2000, Documents,
p. 81.)

The dimensions of this piece, in Anglo-American points:

typeface: Monotype Gill Sans 262, 10 on 16 pts
capital height: 8 pts (to nearest point)
x-height: 6 pts
capital height : line-increment ratio is 8 : 16 pts (1 : 2)
line length (measure) of main text: 200 pts [12 1⁄2 × 16]
text depth: 25 lines × 16 pts = 400 pts
text area is thus in the ratio 1 : 2 (a root 5 rectangle)
page size: 140 × 220 mm
modulus = 16 × 16 pts (see figure 61, overleaf)

Although Froshaug uses the Anglo-American point as a sub-

modular interval, his modulus is the square of the line-increment,
as was true of Aldus. He also centres punctuation marks in the
space between words. Here the space appears to be 2 × word
space, or:
hair space | punctuation mark | hair space

In common with most lettering artists who designed typefaces,

Eric Gill seems not to have been called upon by the type manufac-
turer to quantify such variables as capital height and lowercase
vowel height (x-height) relative to the pk-heights of the character
set. In this piece, Froshaug follows the standard established by
Aldus for a line-increment of twice the capital height.

Opposite: 60 ‘Initium sancti evangelii secundum Joannem’

Overleaf: 61 ‘Initium sancti’: modulus

132
100 %
62 Ovid, Metamorphoses

The ratio of capital-height to line-

increment (1 : 2) is the same in the
Froshaug and the Aldus page. It pro-
vides sufficient space between lines
to permit the rational spacing of all-
capital headings. Although Aldus’s
page is ‘set solid’ (no leading), the Gill
typeface of the Froshaug piece requires
6 points of leading to achieve the same
end.

Type spaces 135

Supplement 3
Type size: a system of dimensional references
Ernest Hoch and Maurice Goldring
The project for the international unification of typographic
measurements was launched at the first International Congress
of Icograda in 1964. Its purpose was to overcome the confusion
inherent in the simultaneous use in the printing industry of at
least four conflicting systems of mensuration: inch, centimetre,
Anglo-American point, and Didot point.1,2 In the course of the
work it seemed necessary not only to strive for agreement on the
consistent use of one system of mensuration, but to develop a
system of dimensional references permitting precise description
of type size.
It is widely recognized that the traditional method is unsatis-
factory; the method describes type of widely varying visual size by
uniform reference to body size, with perhaps the addition of loose
terms like ‘condensed’, ‘normal’, ‘extended’, ‘small face’, ‘titling’
etc. Zachrisson, for instance, postulates attention to ‘visual size’
rather than body size, and relates this in the first place to x-height,
in the second place to l-height. He considers variations in the capi-
tal height not large enough to warrant a special standard measure,
and states that descender length does not seem to be of essential
importance. ‘Visual size’ is defined by ‘x-height measured in mm
multiplied by the mean width of the letters’.3 At least one compos-
ing service working to metric dimensions uses ascender height as
the criterion.4
The authors set out to evolve a comprehensive system of
dimensions to which precise values could be ascribed. The
absolute values of these dimensions, together with their propor-
tions, were to approximate a description of any particular type
size, independent of the terminology used in any particular
language. It was taken as a prime requirement that the system
should not be tied to any specific phase of printing technology;
for instance, it should not be restricted to either metal or film but
be applicable to both.
It was a further consideration that the step from precise de-
scription to specification should be considered: potentially, the
136
Reprinted from the text published in:
Typographica, new series, no. 13, 1966.
1 ordination of work in this field. The results are embodied in three
Ernest Hoch, Towards a unified system of
typographic measurements. A paper
read to the International Congress of
Icograda, Zurich, 1964.
2 plete in this respect, and that at least one basic detail diagram is
Aims and terms of reference of the United
Kingdom working party of the Icograda
Project, London, 1965.
3 y1 line to line distance
Bror Zachrisson, Studies in the legibility
of printed text, Stockholm: Almqvist &
Wiksell, 1965.
4
Studio Hollenstein sa, Paris.
use of computing devices would allow latitudes in specification
that could not be realized or even envisaged in the pre-computer
period. In this context, an agreed system of dimensions would be
necessary for specification as well as for critical evaluation by
typographers. It was postulated that the system of dimensions
should form a framework permitting restricted use of only that
part applying to each particular situation, as well as extension and
further development of any particular aspect. In practical terms,
the system must not be too cumbersome for reference in an every-
day context, and it must on the other hand provide for an easy and
systematic classification allowing for future developments.
The results of this work to date are offered for critical comment
and discussion. It is hoped that this will stimulate related research
and investigations and encourage the exchange of ideas and co-
diagrams (figures 1, 2 and 3), which deal with vertical dimensions
only. Work on the horizontal dimensions has not yet reached
publication stage. We are aware that the main diagram is incom-
missing.
Figure 1
The following dimensions cover the essential elements required to
describe vertical type size:
x1 x-height
x2 ascender height
x3 height including ascender and descender
y2 body height (metal)
Type spaces 137
Figure 2
The general relation between the two dimensions is y1 = y2 + Δy.
In terms of metal, it expresses the amount of leading. In terms of
film, it refers to the ratchet arrangements on the drum driving unit
of the Monophoto (at present graduated in half-point steps), to
the film advance mechanism on the Linofilm (graduated in one-
point steps, with fractional changes possible manually) etc.
Condition 1: Δy = 0 or y1 = y2
In terms of metal, no increment means solid setting. In terms of
film, the equation is either a tautology or expresses that the film is
advanced by an amount corresponding to the body size of an
equivalent metal fount.
Condition 2: Δy > 0
In terms of metal, the increment denotes the amount of leading,
or the amount by which the body size is larger than the size used
for solid setting.
In terms of film, y2 has only a derived meaning. (Any attempt to
base definitions in the context of photocomposition on terminolo-
gy taken over from metal is on a par with phenomena such as the
modelling of early railway carriages on the stage coach.) The
dimension of primary importance can be seen to be y1. In practical
use of a system such as that put forward here, both y1 and y2 would
be referred to in the context of metal composition, whereas the
primary dimension y1 would suffice in the context of photo-
composition.
138
Condition 3: Δy < 0
In terms of metal composition, this is normally not meaningful
(filing off the beard in display work can be disregarded for the pur-
pose of this study).
In terms of photocomposition, this will retain meaning so long
as the convention of expressing type size in terms of body size per-
sists, although this has only limited significance, if any.
Figure 3
Three x-dimensions appear to be essential for description of type
size for languages using the Latin, Greek, and Cyrillic alphabets.
Whereas x1 (x-height) was an obvious choice, two possibilities
were considered for ascenders and three possibilities for the third
dimension to complete the system.
Ascender height could be expressed in relation to x-line or to
base-line; either way, the three relevant vertical data are given by
two dimensions and their sum or difference respectively. lt was
decided to follow convention and relate both x1 and x2 to the base-
line, in order to facilitate direct comparison of ascender height
irrespective of variations in x-height.
The x2 dimension is in the nature of a first approximation.
Where capital height differs from ascender height, distinct values
can be ascribed to them. In further approximation, short ascen-
ders such as t, or punctuation marks above the x-line, accents and
umlauts can be described as x2.1, x2.2 etc. According to variations in
the frequency with which capital letters, accents, and other fea-
tures occur in different languages, such information may vary in
significance.
Finally, x3 describes the extreme vertical dimension. This
dimension is relevant to line spacing and unambiguous, whereas
the relation of x3 to y2 is to some extent arbitrary. Relating x3 to y1
appears to be more satisfactory on logical as well as on practical
grounds.
Type spaces 139
Bibliography
Barker, 1992: Nicolas Barker, Aldus Manutius and the development
of Greek script and type in the fifteenth century, 2nd edn, New
York: Fordham University Press, 1992
Andrew Boag, ‘Typographic measurement: a chronology’,
Typography Papers, no. 1, 1996, pp. 105 – 21
Harry Carter, ‘Optical scale in typefounding’, Typography, no. 4,
1937, pp. 2 – 6
Carter, 1969: Harry Carter, A view of early typography up to about
1600, Oxford: Clarendon Press, 1969
Martin Davies, Aldus Manutius: printer and publisher of Renais-
sance Venice, London: British Library, 1995
Henry George Fletcher, New Aldine studies, San Francisco:
B. M. Rosenthal, 1988
Fournier, 1764: Pierre-Simon Fournier, Manuel typographique
[1764 & 1766], ed. James Mosley, 3 vols, Darmstadt: Technische
Hochschule Darmstadt, 1995
Froshaug, 1964: Anthony Froshaug, Typographic norms,
Birmingham / London: Kynoch Press / DADA, 1964; and in:
Kinross (2000, Documents, pp. 179 – 82)
Anthony Froshaug, ‘Typography is a grid’, The Designer, no. 167,
1967, pp. 4 – 5; and in: Kinross (2000, Documents, pp. 187 – 90)
Gombrich, 1984: E. H. Gombrich, The sense of order, 2nd edn,
London: Phaidon Press, 1984
Ernest Hoch & Maurice Goldring, ‘Type size: a system of dimen-
sional references’, Typographica, new series, no. 13, 1966,
pp. 53 – 6
Allen Hutt, Fournier: the compleat typographer, London: Frederick
Muller, 1972.
Robin Kinross, Modern typography: an essay in critical history,
London: Hyphen Press, 1992
Kinross, 2000: Robin Kinross, (ed.), Anthony Froshaug: typography
and texts / documents of a life, London: Hyphen Press, 2000
Kula, 1986: Witold Kula, Measures and men, Princeton, New
Jersey: Princeton University Press, 1986
Martin Lowry, The world of Aldus Manutius: business and scholar-
ship in Renaissance Venice, Oxford: Basil Blackwell, 1978
Bruce Martin, Standards and building, London: Royal Institute of
British Architects, 1971
May, 1995: Alan May, ‘Making Moxon’s mould’, Printing Historical
Society Bulletin, no. 40, 1995 /6, pp. 15 – 23
140
Morison, 1926: Stanley Morison, Type designs of the past and pre-
sent, London: The Fleuron, 1926
Mosley, 1991: James Mosley, ‘Illustrations of typefounding
engraved for the Description des arts et métiers’ of the
Académie Royale des Sciences, Paris, 1694 to c. 1700’, Matrix,
no. 11, 1991, pp. 60– 80
Mosley, 1997: James Mosley, ‘French academicians and modern
typography: designing new types in the 1690s’, Typography
Papers, no. 2, 1997, pp. 5 – 29
Moxon, 1683 – 4: Joseph Moxon, Mechanick exercises on the whole
art of printing [1683 – 4], 2nd edn, ed. Herbert Davis and Harry
Carter, London: Oxford University Press, 1962.
Osley, 1980: A. S. Osley, Scribes and sources: handbook of the
Chancery hand in the sixteenth century, London: Faber & Faber,
1980
Parkes, 1992: M. B. Parkes, Pause and effect: an introduction to the
history of punctuation in the West, Aldershot: Scolar Press, 1992
Proctor, 1900: Robert Proctor, The printing of Greek in the fifteenth
century, Oxford: Oxford University Press [Bibliographical
Society Illustrated Monographs, no. 8], 1900
Renouard, 1834: A. A. Renouard, Annales de l’imprimerie des Alde,
3rd edn, Paris, 1834
Simpson, 1935: Percy Simpson, Proofreading in the sixteenth, seven-
teenth and eighteenth century, Oxford: Oxford University Press,
1935
Smeijers, 1996: Fred Smeijers, Counterpunch: making type in the
sixteenth century, designing typefaces now, London: Hyphen
Press, 1996
Smith, 1993: Margaret M. Smith, ‘The pre-history of “small caps”:
from all caps to smaller caps to small caps’, Journal of the
Printing Historical Society, no. 22, 1993, pp.79 – 106
Richard Southall, ‘A survey of type design techniques before 1978’,
Typography Papers, no. 2, 1997, pp. 31 – 59
Southward, 1912: John Southward, Modern printing: a handbook,
3rd edn, vol. 1, London: Raithby, Lawrence, 1912
Ullman, 1960: B. L. Ullman, The origin and development of human-
istic script, Rome: Edizioni di Storia e Letteratura, 1960
A. P. Usher, A history of mechanical inventions, 2nd edn,
Cambridge, Massachusetts: Harvard University Press, 1954
F. J. M. Wijnekus, Elsevier’s dictionary of printing and allied trades,
Amsterdam: Elsevier, 1967
Bror Zachrisson, Studies in the legibility of text, Stockholm:
Almqvist & Wiksell, 1965
Type spaces 141
Index
Académie des Sciences, 26n
Aldus Manutius
– books printed by him or successors:
Bembo, De Aetna (1495 /6), 12, 13, 15,
23, 37, 42 – 3, 66, 67, 68
Herodotus (1502), 67
Juvenal and Persius (1501), 86
Lascaris, Erotemata (1495), 12, 13,
14 – 15, 23, 35n, 37, 38 – 41, 67, 68,
71 – 82, 83, 87, 113
Martial (1501), 126 – 8
octavo classics series, 12, 33, 67, 68,
71, 83 – 7, 104 – 5, 108 – 12
Ovid, Metamorphoses (1502), 13,
16 – 17, 33, 34n, 35, 37, 44 – 5, 68, 83,
84, 85, 86, 98 – 9, 101, 102 – 3,
106 – 12, 113, 135
Pliny (1509), 130
Pontano (1505), 129
Seneca, Quaestiones naturales (1522),
9 – 10, 13, 18 – 20, 23, 33, 35, 37,
46 – 65, 68, 101, 113, 114 – 21
Sophocles (1502), 131
– types produced for him, see: Greek-
script types; Latin-script types
Amman, Jost, 10n
ascender-height, 11n, 68, 72, 100, 136,
137, 138
Barker, Nicolas, Aldus Manutius and the
development of Greek script, 72, 86n
base-line, 11, 29, 35, 38, 70, 72, 76, 78,
85, 138, 139
Blunt, Mr, 124n
Burgess, John, 36n, 73n
capital letters, 87, 113, 131
capital-height, 90– 100, 124, 135n, 136,
139
Carter, Harry
A view of early typography, 10n, 69n,
85n
edition of Fournier, 29n, 82n, 122,
123n
edition of Moxon, 26
Caslon, William, 27
Chomsky, Noam, 87n
Davis, Herbert, edition of Moxon, 26
descender-height, 11n, 35, 72, 100, 136,
137
142
Erasmus, Desiderius, 71
Fertel, Martin Dominique, La science
pratique de l’imprimerie, 26
Fournier, Pierre-Simon, 37, 67, 124
‘Imprimerie en caractères’, 26
Manuel typographique, 11n, 26, 27, 29,
82n, 122 – 3
Froshaug, Anthony, 124
homage to Aldus, 73n, 132 – 5
Typographic norms, 25
‘Typography is a grid’, 8
gauges for sizing type
‘face gauge’, 26 – 7, 28, 29, 37, 66
spaces used in tooling up, 66, 67, 68
Gill, Eric, 132
Gill Sans, 132, 135n
Goldring, Maurice, 101n, 136
Gombrich, Ernst, 87
Greek-script types of Aldus, 9, 11, 67
first, 12, 38 – 41, 65, 67, 68, 69, 70,
71 – 82, 90, 91, 100
second, 12, 65, 67, 68, 69, 83, 85 – 6,
92, 100
third, 12, 67, 68, 69, 70, 86, 94, 100
fourth, 12, 33, 67, 68, 69, 70, 82n, 86,
95, 97, 100, 131
grids, reference, 34, 101 – 21
Griffo, Francesco (of Bologna), 9, 11, 12,
27, 33, 66, 67, 72, 85, 97, 102 – 3, 124
Gutenberg, Johann, 10
Hackelsberger, Angela, 101n
Haebler, Konrad, 72n
Hansard, T. C, Typographia, 11n
Henze, Adolph, Handbuch der
Schriftgießery, 122
Hoch, Ernest, 136
Towards a unified system of typo-
graphic measurement, 136n
Icograda (International Congress of
Graphic Design Asociations), 136
inter-x-height space, 22
isomorphism in typography, 9, 11, 124
italic, see: Latin-script types of Aldus
Jacobi, C. T, Printing, 26
Jenson, Nicholas, 72
Johnson, John, Typographia, 26
justification of text, 26, 35, 49, 124, 125
kerning, 38, 73, 76, 78
Kula, Witold, Measures and men, 8
Latin-script types of Aldus
for Bembo edition, 12, 42 – 3, 65, 68,
69, 70, 93, 100
for Lascaris edition, 38 – 41, 65, 67, 68,
69, 70, 87
capitals: octavo classics, 12, 33
cursives: octavo classics, 9, 11, 33,
44 – 65, 67, 68, 69, 70, 85, 96, 97
leading, 11
Leto, Pompiano, 33
line-increment, 11, 22, 23, 66, 70, 73,
124, 137 – 9
Linofilm, 138
Mardersteig, Giovanni, 72n
May, Alan, 10n, 11n, 12n, 36n, 73n, 122
‘Making Moxon’s mould’, 36n
measurement, method of, in this book,
22 – 3
measurement in typography, see:
norms of measurement in typo-
graphy
Mercator, Gerardus, Literarum
Latinarum, 27, 28
Monophoto, 138
Morison, Stanley, Type designs of the
past and present, 85n
Mosley, James, 11n, 72n
‘Illustrations of typefounding’, 26n
‘French academicians and modern
typography’, 26n
edition of Fournier, 122, 123n
mould, typefounder’s, 10– 11, 12, 27, 36,
37, 122 – 3, 124
mould size, see: line-increment, type
size
Moxon, Joseph, Mechanick exercises, 26,
27, 28, 124
Nelson, Stan, 10n
Niccoli, Niccolò, 85
norms of measurement in typography,
10, 12, 25, 136; see also: type size
Aldine, 12, 22, 23, 33, 35, 36, 67, 87, 124
norms of measurement in typography vowel height, see: x-height
(continued)
Anglo-American, 25, 34 – 5, 36, 132, watermarks, 114, 118
136 word-spacing, close, 35, 79
Didot system, 25, 136 writing, 9, 10n, 25, 33, 34, 71, 72, 8, 125
Fournier’s system, 67
x-height, 22, 27, 43, 57, 66, 68, 72, 78, 79,
Osley, A. S, Scribes and sources, 10n, 27, 85, 90– 100, 124, 136 – 9
30
Yciar, Juan de, 10n
Parkes, M.B, Pause and effect, 112n
Planck, Stefano, 72n Zachrisson, Bror, Studies in the legibility
Plantin, Christophe, 11n of printed text, 136n
Popper, Karl, 87
Proctor, Robert, The printing of Greek in
the fifteenth century, 71, 72, 73
punchcutting, 71
punctuation marks, 52, 112, 130
Renouard, A.A, Annales de l’imprimerie
des Alde, 115n
serifs, 66
Simpson, Percy, Proofreading in the
sixteenth, seventeenth and eighteenth
century, 34n
Smeijers, Fred, Counterpunch, 12n
Smith, Margaret M, ‘The pre-history of
“small caps” ’, 113n
Soncino, Gershom, 72n
Southward, John, Modern printing,
25 – 6, 35
spaces: passim, and see: line-increment;
norms of measurement in typo-
graphy; word-spacing
Stafford College of Art and Design, 36n,
72n
structure in typography, 9, 26, 27, 34, 87
Studio Hollenstein, Paris, 136n
symmetry, 125, 126, 127
Thym, Moses, 34n
Times Roman, 85n
Torresani, Andrea, 12, 33, 72
type size, 11, 37, 136 – 9; see also: norms
of measurement in typography
system of named type sizes, 12, 67, 69,
71
Typographica, 136n
Ullman, B. L, The origin and develop-
ment of humanistic script, 8
Vernier, P, 12n
Vingles, Juan de, 10n
Type spaces 143
Postscript [2019]

Peter Burnhill’s Type spaces was published in October 2003 in an Bibliograpical Society / Printing Historical Society, 2016) used
edition of 1,000 copies. By 2011 the last copy had been sold. While Burnhill’s work and its method of measuring risen spaces, in her
the book was no doubt the most specialized title to be published by discussion of the measuring system that Zainer may have used.
Hyphen Press, its implications and suggestions are wide-ranging. Her work is another step towards confirming that early printers
It is reissued in this form as a pdf file, not least for new readers did develop and implement systems of regulating the dimensions
coming to typography in recent years. No significant changes have of their materials.
been made to the text and pictures published in the book of 2003.
The published discussion that has surrounded this work can be Peter Burnhill died in March 2007. Obituaries of him by Paul Stiff
simply resumed. may be found here:
In the years in which Peter Burnhill worked towards the book theguardian.com/news/2007/jun/22/guardianobituaries.obituaries
that was published as Type spaces (the working title had been optimism-modernity.org.uk/people/burnhill.html
‘Griffo’s grid’) he published a short exposition of his ideas in My own memories of him are on the Hyphen Press website:
Typography papers 4 (Reading: Department of Typography & hyphenpress.co.uk/journal/article/remembering_peter_burnhill
Graphic Communication, University of Reading, 2000). I co-edit-
ed that issue of Typography papers, and I hoped that pre-publica- Robin Kinross
tion would help Peter to refine his ideas. We invited and published London, March 2019
comments by Richard Southall, Peter Enneson, Andrew Boag,
Hrant Papazian, and Paul Stiff; and then concluded the sympo-
sium with Peter’s response.
When eventually published, Type spaces was not much noticed
in journals or magazines. Phil Baines reviewed the book sympa-
thetically for Eye (no. 52, 2004). Margaret M. Smith outlined a
printing historian’s reservations in The Library (7.5.3, September
2004). Generous appreciations came from France and Germany:
Jacques André in La Lettre Gutenberg (no. 29, February 2005) and
Roland Reuß in Text (Heft 9, 2004). This, and some online discus-
sion, seems to have been the sum of the book’s critical reception.
One piece of empirical testing of Peter Burnhill’s theses was
carried out, around 2005, by Fred Smeijers and his type design
students at the Hochschule für Graphik und Buchkunst, Leipzig.
They looked at the roman and italic types used in examples of
Aldine printing that they could borrow, photograph, and enlarge.
They worked on whole lines of text, rather than the small portions
that Burnhill used. Their analysis confirmed Burnhill’s Aldine
module (1 /12 of an em); though over the length of a line the sys-
tem seemed always to break down. Their work ended inconclu-
sively and was unpublished.
Claire M. Bolton, in her The fifteenth-century printing practices
of Johann Zainer, Ulm, 1473–1478 (Oxford / London: Oxford

144 Type spaces 145